Content uploaded by Spencer G. Lucas
Author content
All content in this area was uploaded by Spencer G. Lucas on Oct 06, 2023
Content may be subject to copyright.
559
Lucas et al., 2023, Fossil Record 9. New Mexico Museum of Natural History and Science Bulletin 94.
LATERAL SKULL ANGLE: A NEW SEXUAL DIMORPHISM SIGNAL IN
TEMNOSPONDYL AMPHIBIANS
LARRY F. RINEHART and SPENCER G. LUCAS
INTRODUCTION
Sexual dimorphism is common in vertebrates and often
indicates not only gender, but reproductive tness to potential
mates (Fairbairn et al., 2007). Dening the types and extent
of sexual dimorphism and understanding its evolution are
important in discovering the demographics, and predicting
likely behaviors in, extinct and extant populations (Fairbairn, et
al., 2007). For example, the degree of sexual size dimorphism
in a species is often associated with behaviors in pair bonding,
monogamy, paternal care of juveniles, mate selectiveness, and
the extent of male-male aggression (Huxley, 1966). Assignment
of these behavioral traits to a fossil population must, of course,
be made carefully, and hopefully with other corroborative
evidence. Such was the case in Coelophysis bauri, a Late Triassic
theropod dinosaur. Rinehart et al. (2009) found a distinct, but
low value sexual size dimorphism index (e.g., Fairbairn, 1997)
in C. bauri that may be associated with, among other behaviors,
parental care of the juveniles (Huxley, 1966). The survivorship
curve for the same Coelophysis population showed extremely
low mortality in juveniles less than one year old after which
the mortality rate increased dramatically. The parsimonious
interpretation of these two disparate facts seems to be that the
animals provided parental care and/or protection for one year
(presumably until the next breeding season).
Interpreting whether various dimorphic structures present
in fossil populations represent sexual dimorphs, mixed species,
or even pathologies is fraught with problems and controversy
(e.g., Mallon, 2017). This is particularly true when only single
specimens or small population samples are available for study.
An excellent example may be found among the Late Triassic
phytosaurs, an extinct clade of archosaurians similar in shape,
habitat, and lifestyle to crocodilians. Two supposed phytosaur
species, Pseudopalatus pristinus and Pseudopalatus buceros,
were recognized as almost certainly sexual dimorphs of the
same species, P. buceros (Zeigler, et al., 2002, 2003, g. 9 and
associated text). The skulls of these two coeval and sympatric
“species” were essentially identical except for a large narial crest
in P. buceros, the presumed male, and a more gracile rostrum in
P. pristinus, the presumed female.
Abstract—Cranial sexual dimorphism is common among vertebrates. The array of antlers, crests, beards,
wattles, and other sexual display structures is vast, not to mention the many size, shape and color variants
that may apply to all parts of the body. Here, we explore another possible sexual dimorphism signal seen
in the skulls of at least some temnospondyl amphibians: the angle between the straightest portions of
the lateral skull margins, dubbed the lateral skull angle. We investigate this phenomenon principally in
the Metoposauridae but show that it exists in some, though not all, other temnospondyl taxa. The lateral
skull angle was measured between the straightest portions of the lateral skull margins, typically parallel
to the maxillary toothrows. Statistical tests show that the distribution of lateral skull angles is bimodal,
comprising distinct wide and narrow angle variants. We performed additional tests to show that within
taxa that show dimorphism, the lateral skull angle is independent of ontogenetic growth. The statistical
tests dene the lateral skull angles and relative proportions of the two morphs. Where sucient fossil
material was available, we tested other bones, particularly those of the pelvis, and showed that they often
have two shape variants that occur in proportions similar to those of the lateral skull angle. The above
correlations suggest that the bimodal nature of the lateral skull angle may represent sexual dimorphism
in the shape of the face and we demonstrate a high probability that they do. If the observed dimorphisms
are sexual in nature, they make possible the determination of sex ratios in these extinct animals—a
signicant ecological and ethological factor.
New Mexico Museum of Natural History and Science, 1801 Mountain Road, Albuquerque, New Mexico 87104; larry.rinehart@earthlink.net
Sexual dimorphism in humans, including the shape of
the face, is well studied (e.g., Geary, 1998). For example, the
inclination of the forehead is distinctly greater in males than
in females. Petaros et al. (2017) found frontal bone inclination
measured from the glabella to vary between 73o and 78o in
males, and between 80o and 83o in females. This relatively small
dierence of 6o on average is, however, perfectly obvious to
the human eye. Additionally, among many other characters, the
male face is longer and more rectangular than the female face,
which is squarer (Ferrario et al., 1993).
In some cases, characters that have been represented as
sexually dimorphic do not stand up to statistical testing. For
example, the lateral angle of the temporal bone was proposed
and often cited as a sexually dimorphic character to be used
in the forensic analysis of human remains (Wahl, 1981). In
this case, the lateral angle was dened as “the angle between
the lateral wall of the internal acoustic meatus and the
medial surface of the petrious portion of the temporal bone.”
Subsequent investigations, however, concluded that the lateral
angle character was not diagnostic, the accuracy rate being only
53 percent (Bonczarowska et al., 2021). The temporal bone does
remain important in sex assignment of human remains, but it is
the robustness of the mastoid process (signicantly greater in
males), not the lateral angle that is pertinent (Kozerska et al.,
2015).
The Temnospondyli (Zittel, 1887-90) were an order-
level taxon of Paleozoic amphibians that survived well into
the Mesozoic. Comprising nearly 300 species, they were the
most speciose clade of Paleozoic tetrapods and are considered
probably ancestral to the extant lissamphibians (Schoch,
2014). The temnospondyls are dened by the structure of their
vertebrae, their labyrinthodont teeth, a at skull with palatal
vacuities, stapes morphology, and several postcranial skeletal
features (Schoch, 2014).
In the following sections we investigate variation in
the lateral skull angle, as dened below (Fig. 1A), of some
temnospondyl amphibians. The study animals include the
metoposaurs Koskinonodon perfectum, Dutuitosaurus ouazzoui,
and Metoposaurus diagnosticus, the capitosaur Eocyclotosaurus
560
561
appetolatus, and the actinodontid, Sclerocephalus haeuseri
(Fig. 1B-D). We detail the statistical testing, dene the
distribution of lateral skull angles and compare results between
various temnospondyl taxa. We show evidence that the most
parsimonious interpretation of the two morphological variants
we identify is that they are sexual dimorphs.
Institutional Abreviations: The following abbreviations
are used here: MCZ, Museum of Comparative Zoology,
Harvard, Cambridge, Massachusetts; MNHN, Muséum
National D’Histoire Naturelle, Paris, France; NMMNH, New
Mexico Museum of Natural History and Science, Albuquerque,
New Mexico; NMNH, National Museum of Natural History,
Smithsonian Institute, Washington, D.C.; PPHM, Panhandle
Plains Historical Museum, Canyon, Texas; SMNS, Staatliches
Museum für Naturkunde Stuttgart, Stuttgart, Germany; UGKU,
Urweltmuseum Geoskop, Thallichtenberg, Germany; YPM,
Yale Peabody Museum of Natural History, New Haven,
Connecticut; ZPAL, Institute of Paleobiology, Polish Academy
of Sciences, Warsaw.
PREVIOUS OBSERVATIONS ON SEXUAL
DIMORPHISM IN TEMNOSPONDYLS
In most of the literature where sexual dimorphism is
addressed in temnospondyls, the authors conclude that observed
variants or dimorphisms may be, but probably are not, sexual
in nature. In no case that we are aware of, including the
present study, do the authors consider sexual dimorphism in
temnospondyls to be proven. Damiani (2008) states atly that,
“…clear evidence of sexual dimorphism in temnospondyls
has yet to be recognized.” We provide here a sampling of
observations of possible sexual dimorphism in temnospondyls.
They are presented in chronological order of publication.
Australerpeton cosgri —In their study of the squamation
of Australerpeton cosgri, Dias and Richter (2002) noted
variations in size and small dierences in the postcranial
skeletons of their study specimens. They concluded that sexual
dimorphism was a possible cause, but that ontogenetic variation
was more likely.
Xenotosuchus africanus—Damiani (2008) discussed
the possibility that a giant skull of Xenotosuchus africanus,
a Middle Triassic mastodonsaurid, could represent a sexual
dimorph because of its great size, but thought the specimen
probably represented an ontogenetic extension of the known
growth series for the species.
Gerrothorax pulcherrimus—The histological study of
Sanchez and Schoch (2013) showed that some G. pulcherrimus
specimens took twice as long to reach sexual maturity as others.
They considered the idea that this could result from sexual
dimorphism but rejected this in favor of ecological factors.
Milnererpeton huberi—Werneburg et al. (2013) discovered
a possible sexually dimorphic character in the branchiosaur-like
amphibian, Milnererpeton huberi: the mineralization, or not, of
the caudal notochord covering. They made a strong case that
this character was sexually dimorphic based on comparison to
extant salamander larvae. In these larvae the mineralization of
the notochord covering in males leads to a stronger tail region,
which, in turn, provides advantages in swimming, mating, and
defense.
Metoposaurus—The clavicle to interclavicle articulation
in M. diagnosticus krasiejowensis is highly variable (using
the taxonomy of Sulej, 2002). Some individuals show a
straight-line contact whereas others show a protuberance on
the posteromedian edge of the clavicle and a corresponding
accommodation in the interclavicle (Sulej, 2002). Sulej (2002)
states that this variation cannot represent sexual dimorphism
because there are intermediate forms between these two end
points. In this regard, we point out that sexually dimorphic
characters exist on a statistical continuum, and therefore
intermediate forms are to be expected. Individual expression of
sexually dimorphic characters is highly variable. For example,
sexual size dimorphism is well understood in humans where the
men are statistically taller, but the tallest women are taller than
the shortest men (Kirchengast, 2014, citing Holden and Mace,
1999). The distinct statistical distributions of male and female
heights overlap considerably. The existence of intermediate
forms does not rule out sexual dimorphism.
Antczak and Bodzioch (2018) found two distinct
ornamentation patterns in dermal bones of M. krasiejowensis
(taxonomy of Brusatte et al., 2015). These patterns comprised
(1) ne, regular, and sparse and (2) coarse, irregular, and dense
morphotypes. Taxonomic, ecological, individual or ontogenetic
variation, and sexual dimorphism were proposed as possible
explanations. Ultimately, they determined that ecological factors
or the existence of a neotenic population were more likely
explanations than ontogenetic or sexual dimorphism factors.
Teschner et al. (2017), through histological study of humeri,
discovered variable growth patterns in M. krasiejowensis
(taxonomy of Brusatte et al., 2015). Two histotypes were found
to dier in vascularization and degree of remodeling. They
proposed temporally distinct populations that may have been
mixed by reworking, sexual dimorphism, and random variation
as explanations. Their conclusion was to continue the work,
concentrating on geochemical analysis to test the hypotheses.
Onchiodon labyrinthicus—Schoch (2021) noted an
enhanced degree of morphological variation in O. labyrinthicus
compared to some other temnospondyls (e.g., Sclerocephalus)
which, among other factors, could be attributed to sexual
dimorphism. He concluded, however, that sexual dimorphism
was less likely to account for the excessive variation than genetic
factors or plastic response to a changing environment.
METHODS
In the temnospondyls under study here, there exists a
distinctly straight segment of the maxillae. This essentially
straight part of the skull margin extends from the ventromedial
process of the maxillae, which enters the choanae, posteriorly
for all, or nearly all, the length of the maxillary toothrow. For
examples of this morphology in Koskinonodon perfectum, see
Lucas et al. (2016, gs. 27-34); in Dutuitosaurus ouazzoui, see
Dutuit (1976, gs. 2, 3); in Metoposaurus diagnosticus, see Sulej
(2007, gs. 4-9 and 11-15); in Eocyclotosaurus appetolatus, see
Rinehart and Lucas (2016, gs. 20-31); and in Sclerocephalus
haeuseri, see Schoch and Witzmann (2009, g.4). For the
purposes of this study, the lateral skull angle is dened as the
angle between the skull margins along these straight-line or
essentially straight-line portions of the maxillary toothrows (Fig.
1A). We note that this metric can be measured on incomplete
skulls as long as a reasonable segment of the maxillary toothrow
and the midline suture are preserved.
Measurements were made from photographs of the skulls
FIGURE 1 (facing page). Lateral skull angle dened and life reconstructions of the study temnospondyl amphibians. A, A bone
map of the skull of Koskinonodon perfectum showing the sutures (black) and the lateral line canals (gray). The lateral skull angle
is dened as the angle between the essentially straight segments of the lateral margins of the maxillae. B, Koskinonodon perfectum
by Mathew Celeskey was typical of the three Late Triassic metoposaurs analyzed here. C, Eocyclotosaurus appetolatus by Fredrik
Spindler was a Middle Triassic capitosaur. D, Sclerocephalus haeuseri by Dmitry Bogdanov was a Permian actinodontid amphibian.
Scale approximate for large adult animals.
562
Lateral Skull Angle in Koskinonodon
The same bimodal distribution of lateral skull angles exists
in both the Lamy and Rotten Hill Koskinonodon populations
as indicated by the same inection point in their respective
curves, and the similar mean values and standard deviations
evidenced by the similar slopes of their datapoint lines (Fig.
2A-B). Therefore, we conclude that the observed bimodality is
a character of Koskinonodon and is not locality dependent. We
thus combined the two populations to obtain a larger statistical
sample (N = 26).
A histogram and probability plot of the lateral skull angle
of the combined Lamy and Rotten Hill populations shows
a more distinctly bimodal histogram and two better dened
components that overlap at a distinct inection in the datapoint
line (Fig. 2C). The component distributions were separated at
the inection point and replotted (Fig. 2D). Their respective
means and standard deviations were calculated and tabulated
(Fig. 2D; Table 1). Representative skulls of the two variants are
illustrated (Fig. 3).
The mean (μ) lateral skull angle of the low angle variant
with a broader rostrum is 25.2o, and the high angle variant with
a more acute snout has a mean lateral skull angle of 32.2o (Table
1). The position of the inection point of the datapoint line,
which occurs at the 0.35 probability level (Fig. 2C), indicates
the percentage of the population that belongs to each class
(King, 1971); the low lateral skull angle class contains 35% of
the population, and the high angle class contains 65% (Table 1).
The standard deviation (σ) of these component modes is a
measure of variation (variance = σ2). The standard deviation,
as indicated by the slopes of the datapoint lines, is 1.05o for the
low lateral skull angle class, whereas the standard deviation of
the high angle class is 2.81o (Fig. 2D; Table 1).
Coecient of variation—To accurately compare variation
in the two morphs, we use the coecient of variation (CV)
(Simpson et al., 1960; Van Valen, 2005). This metric, dened as
CV = 100.σ/μ,
removes the eect of size from the comparison.
The CV of the low angle skulls is 4.17, whereas that of the
high angle skulls is 8.73 (Table 1). The high angle skulls show
more than twice the coecient of variation of the low angle
skulls in agreement with the standard deviation above.
Sexual dimorphism?
We have shown that the lateral skull angle in Koskinonodon
does not vary smoothly from a broad snout to a more triangular
one. There are two well-dened variants whose distributions are
normal (Gaussian) and overlap somewhat. But, does this indicate
sexual dimorphism? Alternative explanations for the bimodality
could include mixed species, ontogenetic stages, or other.
Mixed species?—There are no apparent skeletal indications,
such as a dierent arrangement of the skull bones, between the
two variants to indicate the presence of mixed species. The 7o
dierence between the mean values of the high and low lateral
skull angle groups does not seem great enough to indicate a
signicantly dierent feeding style or occupation of a dierent
ecological niche, and certainly not great enough to erect a new
species. Additionally, Lucas et al. (2010, 2016) showed evidence
that both the Lamy and Rotten Hill Koskinonodon were breeding
populations, and, if this is correct, the mixing in of a dierent
species would be most unlikely.
Ontogenetic stages?—Clack (2012), citing Olson
(1951) and Rinehart and Lucas (2001), discussed the sudden
morphological changes in the skull of the Permian nectridean,
Diplocaulus magnicornus, that could be due to metamorphosis
or lifestyle changes. These changes have their onset at a specic
ontogenetic stage and were clearly size-related. They bring
up the question of whether the lateral skull angle variants of
in either dorsal or ventral view by a General Tools® Number
823 25 cm digital angle nder/ruler with approximately 1/10o
resolution.
LATERAL SKULL ANGLE DIMORPHISM IN
METOPOSAURIDS
The Metoposauridae
The Metoposauridae was a family of Late Triassic
amphibians. The adults were typically two to three meters in
length with a laterally compressed tail for swimming, small
limbs that were inadequate for terrestrial locomotion (at least
in the adults), a robust shoulder girdle, and a broad, attened
skull with the orbits in the anterior half (e.g., Colbert and
Imbrie, 1956; Dutuit, 1976; Hunt, 1993; Schoch and Milner,
2000) (Fig. 1B). Metoposaurs are known from North America,
India, Europe, northwestern Africa, and Madagascar (Schoch
and Milner, 2000). Koskinonodon, the metoposaur of greatest
interest here because it is represented by the largest statistical
sample, is a monospecic genus within the Metoposauridae
(Schoch and Milner, 2000), K. perfectum being the only species,
and is endemic to the American Southwest. Here, we follow the
taxonomy of Schoch and Milner (2000).
Koskinonodon
While studying the NMMNH collection from the Lamy
amphibian quarry (see e.g., Romer, 1939; Lucas et al., 2010)
collection of Koskinonodon we observed that some skulls
appear to have a more broadly rounded rostrum and more nearly
parallel sides, whereas others are more triangular with a more
acute rostrum and a greater angle between the lateral margins. A
quick check of photos of the PPHM collection of Koskinonodon
skulls from Rotten Hill, Texas (see Lucas et al., 2016) showed
the same variation in rostral shapes.
We then undertook to dene the nature of this distribution of
morphologies; to determine whether it was a single continuously
varying distribution or composed of distinct variant components.
Skulls that were obviously attened (e.g., skull roof concave or
extreme lateral bulging of the quadratojugal area) or otherwise
distorted (e.g., signicant asymmetry about the midline or
broken into pieces with signicant gaps between them) were
disqualied from the study. Statistical testing, using JMP 10®
(2012) and KaleidaGraph® (2010) statistics and graphical
analysis software, showed that two distinct variants were present,
each distributed normally and with the skirts of their bell-shaped
normal (Gaussian) distributions overlapping. Metrics for the
skulls used in the study are tabulated (Appendix 1).
Lamy population—A histogram and probability plot of the
lateral skull angles of 16 Lamy amphibian quarry Koskinonodon
skulls show a distinct bimodal distribution (Fig. 2A). Although
the histogram has a less clearly bimodal appearance because the
data are binned, the probability plot, a more sensitive test, is
distinctly bimodal as indicated by the long, stretched-S shape
of the datapoint line (King, 1971; Sinclair, 1976; Rinehart et al.,
2022). Thus, the overall distribution of lateral skull angles in this
sample comprises two component distributions. The inection
point of the datapoint line is at approximately 0.35 (35%) on the
probability scale, indicating that 35% of the skulls are members
of the low lateral skull angle component, and 65% of them occur
in the wide angle component (King, 1971).
Rotten Hill population—A sample of 10 Koskinonodon
skulls from the Rotten Hill, Texas, deposit shows a distribution
of lateral skull angle values similar to the Lamy sample (Fig. 2B).
Whereas the histogram is only vaguely bimodal, the probability
plot is clearly so. Again, the inection point of the datapoint line
shows that the division between the two component modes is at
the 0.35 probability level; 35% of the skulls are in the low angle
component, and 65% are in the high angle component as seen in
the Lamy population above.
563
FIGURE 2. Lateral skull angle distribution in Koskinonodon from the Lamy amphibian quarry, New Mexico and Rotten Hill, Texas.
Arrows mark the inection points of the datapoint lines that dene the two component distributions. A, Histogram and probability
plot of lateral skull angles in the Lamy population showing bimodal distribution (N = 16). B, Histogram and probability plot of
lateral skull angles in the Rotten Hill population showing similar bimodal distribution (N = 10). C, Histogram and probability plot
of the combined Lamy and Rotten Hill populations showing better dened bimodal distribution (N = 26). D, Resolved component
distributions of lateral skull angle in Koskinonodon. E, Scatterplot of lateral skull angle against midline skull length showing that
lateral skull angle does not correlate to size (note extremely low R2 numbers).
564
Koskinonodon represent a life stage change.
We answer this question by comparing the lateral skull
angle in Koskinonodon to the midline skull length, which is
used as a measure of size (Fig. 2E). A scatterplot shows that
there is essentially zero correlation between size (as a proxy for
age) and lateral skull angle. R2, the correlation coecient for the
curve ts in the plot is 0.008 for the Lamy population and 3x10-6
for the Rotten Hill group. These extraordinarily low R2 values
show that the lateral skull angle of Koskinonodon is completely
independent of size in the Lamy and Rotten Hill populations,
which were composed entirely of adult individuals. The lateral
skull angles of juvenile Koskinonodon could not be assessed due
to the paucity of specimens.
We therefore consider the hypotheses of taxonomic or
ontogenetic variation to be falsied by this analysis.
Dimorphism in the Postcranial Bones of Koskinonodon
Here, we determine whether dimorphism exists in some
postcranial bones of Koskinonodon. The bones selected for
testing were based in part on sample size and on the likelihood
of sexual dimorphism being present (e.g., the pelvic bones).
Flattened or distorted bones were excluded from the samples.
Clavicle—We tested the width/length (w/l) ratio in a
sample of 14 well-preserved clavicles from the Lamy amphibian
quarry (Appendix 2). The choice of the w/l ratio was intended
as an approximation of shoulder breadth for a given size. A
histogram and probability plot show weakly bimodal behavior
in the distribution of w/l values as indicated by the slight jog
and change of slope (i.e., change in standard deviation) in the
datapoint line (Fig. 4A). The two segments of the bimodal
distribution were separated at the inection point and replotted,
and their reasonably good straight-line ts (note high R2 numbers
in the plot) indicate two normal (gaussian) distributions.
The lower w/l class (narrow morph) mean was 0.46, with a
standard deviation of 0.18, and contains 70% of the population
(Fig. 4A-B). The higher w/l class (wide morph) has a mean w/l
value of 0.52 and standard deviation of 0.22. This wide morph
represents 30% of the population (Fig. 4A-B; Table 1). The CV
of the low and high w/l classes is 3.91 and 4.23, respectively
(Table 1). The high w/l class has greater variation among its
members. Two Koskinonodon clavicles from the NMMNH
Lamy collection (Fig. 4C) have been scaled to the same length
to exemplify the narrow (left) and wide (right) morphs.
The clavicle shape in Koskinonodon is dimorphic, and
the characteristics of the two morphs have been determined.
We performed a nal test to determine if the w/l ratio could
be size-related and thus vary throughout ontogenetic growth.
The correlation coecient (R2) between clavicle length (as a
measure of size) and w/l ratio is 0.058, so there is essentially no
correlation between size and morphology.
Ilium—The ilia of the Lamy amphibian quarry
Koskinonodon population show a distinct posterior oset of the
dorsal process, which occurs abruptly along the anterior edge of
the ilium, usually at approximately 60% of the anterior length
(Fig. 5C). The angle formed between the anterior edge of the
ilium and the oset portion varies from 0o to 27o in a sample of
24 ilia (Fig. 5A-C).
Oddly, the ilia of the Rotten Hill Koskinonodon population
TABLE 1. Summary characteristics of lateral skull angle and postcranial dimorphism. CV = Coecient of variation (Simpson et
al., 1960; Van Valen, 2005).
Taxon Element Angle or w/l
component
N% Population Implicit
sex
ratio
Mean
(μ)
Low mean/
high mean
Standard.
Deviation
(σ)
CV
(100.σ/μ)
Koskinonodon
perfectum
Skull low angle 26 35 1.9:1 25.2 0.78 1.05 4.17
high angle 65 32.2 2.81 8.73
Clavicle low w/l 14 70 0.46 0.018 3.91
high w/l 30 0.52 0.022 4.23
(Lamy only) Ilium low angle 24 42 5.8 2.3 39.7
high angle 58 15.4 4.2 27.3
Ischium low w/l 16 38 0.72 0.013 1.81
high w/l 62 0.84 0.03 3.57
Dutuitosaurus
ouazzoui
skull low angle 13 54 1.2:1 28.4 0.92 0.42 1.48
high angle 46 30.8 0.57 1.85
Metoposaurus
diagnosticus
skull low angle 11 27 2.7:1 29.3 0.9 0.41 1.4
high angle 73 32.4 0.97 2.99
Eocyclotosaurus
appetolatus
skull low angle 14 56 1.3:1 28.3 0.87 0.7 2.47
high angle 44 32.7 1.8 5.5
Sclerocephalus
haeuseri
skull normal dist. 9 100 -- 40.3 6.3 15.6
Homo sapiens frontal low angle (M) 51 1.04:1 75.5 0.93
high angle (F) 49 81.5
565
FIGURE 3. Koskinonodon skulls from the Lamy amphibian quarry, illustrating the two lateral skull angle morphs. These skulls
do not show the extremes of lateral skull angle but are almost exactly at the means of the two lateral skull angle components. A,
NMMNH P-61420, representing the low angle morph with a 25o lateral skull angle, a broader snout and more nearly parallel sides.
B, NMNH 3 (skull number 3 on the NMNH Lamy exhibit block), representing the high angle morph with a 33o lateral skull angle,
a more acute snout, and a more triangular shape.
do not show a similar oset. They have either an approximately
straight anterior edge or a very slight anterior oset that begins
higher on the dorsal process than that seen in the Lamy population
(Lucas et al., 2016, g. 67). This observation is made, however,
on a sample of 4 individuals, which is inadequate to be the basis
of any signicant conclusion.
A histogram and probability plot of the 24 distal ilium angles
from Lamy (Appendix 3) shows weak bimodality as indicated
by the overlapping distributions in the histogram and a small
jog in the datapoint line of the probability plot (Fig. 5). These
indicators signal that there is a large overlap of the skirts of
the two component distributions. However, both the histogram
and probability plot show distinct bimodal behavior in which
a low angle and a high angle mode are present (Fig. 5A). An
arrow marks the inection point between the two component
modes (Fig. 5A), and the separated components are replotted
together with their summary statistics (Fig. 5B; Table 1). As
indicated by the position of the inection, 42% of the population
is contained in the low angle class and 58% in the high angle
class. The standard deviation of the low and high angle morphs,
respectively, is 2.3o and 4.2o, and the corresponding CVs are 39.7
and 27.3. The ilia with a low distal angle show greater variation.
Ischium—In medial or lateral aspect, the shape of
Koskinonodon ischia vary between very lunate and nearly
triangular, wherein they are much wider for a given length (Fig.
6C). We used the width/length (w/l: dened in Fig. 6C) ratio
to dene this shape variation and tested to see if w/l shows a
single continuously varying distribution or if the shape variants
each have their own distribution. A histogram and a probability
plot of 16 ischia measurement ratios (Appendix 4) show that
the distribution of w/l is bimodal, with 38% of the population
falling into the low w/l (lunate) class and 62% belonging to
566
FIGURE 4. Koskinonodon clavicles from the Lamy amphibian quarry show dimorphism in their Width/Length (w/l) ratio. Arrow
marks the inection point of the datapoint line. A, Histogram and probability plot showing bimodal behavior in the w/l ratio (N
= 14). B, The two component modes have been separated and replotted. Component summary statistics are shown in the gure.
Mean = the mean value of the mode, SD = standard deviation, RSQ = R2, the correlation coecient. C, Selected right clavicles of
Koskinonodon are scaled to the same length to exemplify the narrow and wide morphs. Specimens were chosen to be near the mean
w/l ratios of the narrow and wide morphs. The wide morph (left) with higher w/l ratio is NMMNH P-69393, and (right) the narrow
morph with lower w/l ratio is NMMNH P-69685.
567
FIGURE 5. Koskinonodon ilia from the Lamy amphibian quarry show dimorphism in the distal angle of their dorsal processes. A,
Histogram and probability plot of the distal angles of Koskinonodon ilia showing weak bimodality (large overlap of the component
distributions) (N = 24). Arrow shows the inection point where the components were separated. B, Resolved component modes of
the distal ilium angles with their summary statistics. Mean = the mean value of the mode, SD = standard deviation, RSQ = R2, the
correlation coecient. C, Example ilia illustrating the two morphs. MCZ 1925-17, (upper) a high distal angle ilium showing 17.5o
oset of the distal end and MCZ 1925-18, (lower) a low distal angle ilium with 5.5o oset of the distal end.
568
the sub-triangular, high w/l class (Fig. 6A-B). The inection
point indicating the proportions of the component distributions
is indicated by an arrow (Fig. 6A). The resolved low w/l and
high w/l distributions are replotted together with their summary
statistics (Fig. 6B; Table 1). The narrow ischia morph has
a mean w/l of 0.72 and standard deviation of 0.0.13 (CV =
1.81), whereas the wide morph mean is 0.84 with a standard
deviation of 0.3 (CV = 3.57) (Table 1). The wide morph shows
approximately twice as much variation as the narrow morph.
The wider, more triangular component distribution appears
truncated as indicated by the datapoint line bending over and
going horizontal (Fig. 6A-B) (King, 1971). This truncation
onsets at approximately the 0.7 probability level (Fig. 6A). It is
possible that another mode is present that gives the datapoint line
its truncated appearance, but, until a larger sample is available,
this will remain uncertain.
There is a second possible explanation to support the
truncation model. The ischia constitute the lower bony portion
of the pelvic girdle. They are joined by connective tissue
along their ventromedial edges, and they contain a portion of
the acetabulum anterodorsally. Thus, in the animals that have
the high w/l ischial morphology, the acetabula will be farther
apart for a given size (as estimated by length). There may be a
practical limit to the width between the acetabula of a given size
animal before locomotion (including swimming) is negatively
aected. In addition, as the distance between the acetabula
increases, the suspended load between them increases the stress
on the connective tissue, even if the animal is in water and the
limbs are bearing only part of its weight. Either factor, or a
combination could create a practical limit to the width of the
pelvic girdle and cause truncation of the statistical distribution
of the wider morph.
Lateral Skull Angle in Dutuitosaurus ouazzoui
Dutuitosaurus ouazzoui is a metoposaurid amphibian best
represented by the sample from J.-M. Dutuit’s Argana site
13 in the Atlas Mountains of Morocco, where approximately
70 individuals were discovered in what appears to have been
a drying pond or swamp deposit (Dutuit, 1976). The most
unique and important element of this locality is that many of the
animals were fossilized in a fully articulated condition, unlike
the metoposaur bonebeds at Lamy, Rotten Hill, and Krasiejow
(Sulej, 2007; Lucas et al., 2010; Lucas et al., 2016). D. ouazzoui
is similar in bauplan to Koskinonodon but is, on average, a
smaller animal. The principal dierences between the species
are the layout of the skull bones (the lacrimal in particular) and
the shape and ornamentation of the shoulder girdle (Hunt, 1993).
A histogram and a probability plot of 13 lateral skull angle
measurements from the MNHN Dutuitosaurus collection
(Appendix 5) show bimodal behavior (Fig. 7A). As indicated
by the inection point of the datapoint line (arrow in the gure),
the low angle morphs comprise 54% of the population, and
46% belong to the high angle variant (Fig. 7A; Table 1). The
low lateral skull angle mean is 28.4o with a standard deviation
of 0.42o (CV = 1.48), and the mean of the high angle morph
is 30.8o with a standard deviation of 0.57o (CV = 1.85). The
Dutuitosaurus skulls that belong to the high lateral skull angle
morph show greater variation than the low lateral skull angle
morphs, but the disparity is less than that seen between the
Koskinonodon dimorphs (Table 1).
The component distributions of the bimodal distributions
were resolved and replotted (Fig. 7B). Their summary statistics
show a subtle mean dierence of only 2o-3o in the lateral skull
angles of the dimorphs (Fig. 7B; Table 1). Skulls that are
representative of the two morphs were selected, scaled to the
same size, and illustrated (Fig. 7C).
Lateral Skull Angle in Metoposaurus diagnosticus
Metoposaurus diagnosticus is a metoposaurid amphibian
known principally from Germany and Poland, with a particularly
large population sample having been discovered in Krasiejów,
Poland (Sulej, 2007). M. diagnosticus is distinguished from
other metoposaurids by the shape and location of its lacrimal,
which is relatively short and wide, nearly ovate, and does not
enter either the orbit or external naris (Hunt, 1993; Schoch and
Milner, 2000). In addition, the ornamentation and morphology
of the shoulder girdle are diagnostic (Schoch and Milner, 2000).
The lateral skull angles of a sample of 11 skulls, illustrated
by Sulej (2007), was measured. All the study skulls are from
the Krasiejów, Poland population in the ZPAL collections.
These data, together with the associated midline skull lengths
from Sulej (2007, appendix 1), are tabulated (Appendix 6). A
histogram and probability plot of the lateral skull angles shows a
bimodal distribution. The histogram is only weakly bimodal, but
the more sensitive probability plot shows the long, stretched-S
shape distinctly characteristic of bimodality. The position of the
inection in the datapoint line (arrow in Fig. 8A) indicates that
27% of the skulls fall into the low angle component and 73%
belong to the high angle component.
The mean value of the low angle class is 29.3o, and that of
the high angle class is 32.4o, a dierence of only ~3o, and yet the
dimorphism is clear in the plots (Fig. 8A-B) and by observation
(Fig. 8C). As in the Koskinonodon and Dutuitosaurus data,
the standard deviation of the high angle class (0.97o) is much
greater than that of the low angle class (0.41o) (Fig. 8B). The
corresponding CVs are: low angle, 1.4; and high angle, 2.99.
The high lateral skull angle morph shows more than twice as
much variation as the low angle morph.
LATERAL SKULL ANGLE IN OTHER
TEMNOSPONDYLS
Eocyclotosaurus appetolatus
Eocyclotosaurus appetolatus is a capitosaurid amphibian
known from the Middle Triassic Moenkopi Formation of north-
central New Mexico (Rinehart et al., 2015; Rinehart and Lucas,
2016). A single site, the Tecolotito bonebed in San Miguel
County, NM, yielded well over 1700 disarticulated, but well-
preserved bones of the species. Stratigraphic and microfossil
(conchostracans and bivalves) study of the deposit indicate a
likely lacustrine origin (Rinehart and Lucas, 2016).
Eocyclotosaurus appetolatus grew to a maximum of
approximately 2.5 meters in length and possessed a attened
alligator-like skull and a laterally compressed tail adapted to
swimming (Fig. 1C). Functional morphological analysis of the
skulls, jaws, and postcrania point to a generalist feeder in an
alligator-like ecological niche. As in the metoposaurids, the limb
bones were stouter in juveniles and became increasingly more
feeble relative to body mass as the animals grew. The juveniles
probably had terrestrial capability, but the adults almost surely
did not (Rinehart and Lucas, 2016).
Lateral Skull Angle in Eocyclotosaurus appetolatus
The statistical distribution of lateral skull angles in a
population of Eocyclotosaurus appetolatus skulls from the
NMMNH collection (Appendix 7) was tested (N = 14). As in the
metoposaurs, the distribution is bimodal (Fig. 9A). The low angle
morph, with a mean lateral skull angle of 28.1o, encompasses
56% of the population, and the high angle morph, with a mean
of 32.7o, encompasses 44% (Fig. 9B; Table 1). Representative
skulls are illustrated (Fig. 9C).
The standard deviations of the two component distributions
also behave similarly to those of the metoposaurs. The standard
deviation of the low angle morph (0.7o) is less than half that of
569
FIGURE 6. Koskinonodon ischia from the Lamy amphibian quarry show dimorphism in their width/length ratio. A, A histogram and
a probability plot of width/length show two distinct modes (N = 13). The distribution appears truncated on its upper end and may
possibly contain another mode. An arrow marks the inection point separating the two components. B, The resolved component
distributions are replotted and summary statistics provided. Mean = the mean value of the mode, SD = standard deviation, RSQ
= R2, the correlation coecient. C, The measurement protocol for length and width of the ischia is shown (left-center). NMMNH
P-67533, a left ischium (upper), represents the high w/l morph, and MCZ 2013, a left ischium (lower) illustrates the low w/l morph.
570
FIGURE 7. Lateral skull angle distribution of Dutuitosaurus ouazzoui skulls from Argana, Morocco. A, A histogram and a
probability plot show bimodal behavior of lateral skull angles (N = 13). An arrow marks the inection point of the datapoint line
that indicates bimodality. B, The resolved component distributions of lateral skull angles are plotted together with their summary
statistics. Mean = the mean value of the mode, SD = standard deviation, RSQ = R2, the correlation coecient. C, Two skulls from
the MNHN collections, scaled to the same midline length, illustrate the subtle dierence between the (upper) high angle (MNHN
AZA 271) and (lower) low angle (MNHN XIp/4/66) morphs.
571
FIGURE 8. Lateral skull angle in Metoposaurus diagnosticus from Krasiejów, Poland. A, A histogram and a probability plot show
bimodal distribution of the lateral skull angles (N = 11). An arrow marks the inection point of the datapoint line. B, The resolved
component distributions together with their summary statistics. C, Two skulls from the ZPAL collection illustrate the low and high
lateral skull angle morphs. Upper, representing the high angle morph. Lower, representing the low angle morph.
572
FIGURE 9. Eocyclotosaurus appetolatus skulls from the Tecolotito bonebed show dimorphism in their lateral skull angles. A, A
histogram and probability plot of lateral skull angles show bimodal behavior. Arrow marks the inection point of the datapoint line.
B, The resolved component distributions of the low and high lateral skull angle morphs together with their summary statistics. Mean
= the mean value of the mode, SD = standard deviation, RSQ = R2, the correlation coecient. C, Eocyclotosaurus appetolatus skulls
from the Tecolotito, New Mexico bonebed illustrating the two lateral skull angle morphs. Upper, NMMNH P-67401 represents the
high angle class with a lateral skull angle of 35.7o. Lower, NMMNH P-64166 (the holotype) represents the low angle morph with
a 26.4o lateral skull angle.
573
the high angle morph (1.8o) (Fig. 9B; Table 1). The CVs, 2.47 in
the low angle group and 5.5 in the high angle group, show more
than twice as much variation in the high angle morph.
Postcranial Bones of Eocyclotosaurus appetolatus
Ilium—The ilia of Eocyclotosaurus appetolatus show no
oset in their dorsal processes, although they do show a gentle
posterior bend that decreases with size (Rinehart et al., 2016).
This feature of the Lamy metoposaur ilia thus could not be
tested in E. appetolatus.
Ischium—Only six Eocyclotosaurus appetolatus ischia
are available for measurement (Appendix 8). A histogram
and probability plot (Fig. 10A) show a bimodal distribution
in width/length ratios, but the population is too small to make
reliable calculations regarding the proportions of the two
morphs; the addition or subtraction of even a single datapoint
from either distribution could radically change the outcome.
The summary statistics of the two morphs would, likewise, be
unreliable. Nonetheless, the test shows distinct dimorphism in
the ischia of Eocyclotosaurus appetolatus, which is illustrated
by representatives of the two morphs, scaled to the same length
for easier comparison (Fig. 10B). A left ischium, NMMNH
P-67900, represents the low w/l morph with a more lunate shape,
and NMMNH P-62767, a right ischium that has been ipped
horizontally in Photoshop® for easy comparison, represents the
high w/l morph with a wider, more triangular shape (Fig.10B).
Both the wide and narrow morphs are present in large and small
specimens, showing that the shape dierence is not ontogenetic.
Sclerocephalus
Sclerocephalus is a genus of archegosauriform stereospondyl
amphibian within the family Actinodontidae (Schoch and
Milner, 2000) (Fig. 1D). The genus comprises several species,
known from the uppermost Carboniferous through the Permian
of central and eastern Europe. The species of interest here, S.
haeuseri, has a distinctly broad snout, and the lateral margins
of the skull are nearly straight (Schoch and Milner, 2000),
making measurement of the lateral skull angle particularly easy.
Unfortunately, the Sclerocephalus haeuseri study group is not
from a single population; most are from various localities in the
Saar-Nahe Basin of southwestern Germany, and therefore do not
represent a single population.
Sclerocephalus, equipped with a lateral line sensory system
and a laterally compressed tail for sculling, inhabited lakes and
fed primarily on Paramblypterus, a paleoniscoid sh that is
always found in its gut contents (Schoch, 2009; 2014). Trackway
evidence shows that whereas Sclerocephalus occasionally left
the water, it was far from agile on land (Schoch, 2014).
Lateral Skull Angle in Sclerocephalus haeuseri
Here, we present a possible counter example to our proposal
that lateral skull angle dimorphism may be general throughout
the temnospondyls. A growth series of Sclerocephalus haeuseri
skulls varying from 15 mm to 190 mm in midline length was
measured for lateral skull angle (N = 9) (Appendix 9). A
histogram and probability plot of the lateral skull angles does
not show dimorphism (Fig. 11A). Although the histogram
shows possible bimodality, the probability plot shows no
distinct inection. A larger sample would show conclusively
if bimodality were present, but with the available sample,
bimodality cannot be demonstrated; the sample better ts a
single, normal distribution. The R2 correlation coecient of the
data tted to a Gaussian bell curve is 0.952; there is more than
95% correlation between the distribution of the data and a single
perfect Gaussian distribution.
Whereas the lateral skull angle of Koskinonodon was
shown to be independent of size (Fig. 2E), the lateral skull angle
of Sclerocephalus haeuseri shows negative correlation to size,
meaning that the lateral skull angle decreases as size increases
(Simpson et al., 1960) (Fig. 11B). The correlation coecient
(R2 = 0.53) shows that more than half of the lateral skull angle
variation is “explained” by the midline skull length.
The lateral skull angle variation in Sclerocephalus is mostly
ontogenetic; an allometric shape change throughout growth. The
182 mm-long skull of SMNS 90055, the neotype, is illustrated
as an example of a large skull with a low lateral skull angle of
28.9o, whereas SMNS 58784, a medium sized skull of 81mm
midline length, shows a higher lateral skull angle of 38.9o (Fig.
11C).
Although variation in the lateral skull angle of Sclerocephalus
is obvious and veriable, the question of dimorphism remains
unanswered.
DISCUSSION
Observations on Lateral Skull Angles
All three genera of metoposaurs and Eocyclotosaurus
appetolatus show dimorphism in their lateral skull angles (Table
1). In each case, the high lateral angle morphs (more acute snout)
comprise a greater fraction of the population than the low lateral
skull angle morphs (broader snout).
The high angle morphs always show greater variation,
often by a factor of two (CV column, Table 1). The biological
signicance of this, if any, is unknown, and we do not have a
working hypothesis to explain greater variation in one morph
over the other.
Sex Ratio
Assuming that the dimorphism observed here is gender-
based, the sex ratio of the various populations is dened by the
percentage of each population that belongs to each presumed
sexual morph (Implicit sex ratio: Table 1).
Importance and denition of sex ratio—The modeling of
Jennions and Fromhage (2017) has shown that the sex ratio of
a species is dependent upon or can drive important ecological
and ethological factors regarding parental care, including which
gender is the care-giver (e.g., Okada et al., 2014), competition
for mates, and mate selection.
The term sex ratio refers to the relative proportions of males
and females. However, according to Jennions and Fromhage
(2017), distinctly dierent types of sex ratios exist: 1. Sex ratio
at maturity (MSR) is the m/f ratio of a cohort that has reached
maturity. 2. Eective sex ratio (ESR) is the m/f ratio of animals
that mate at least one time. 3. The m/f ratio of living adults is the
adult sex ratio (ASR). 4. Operational sex ratio (OSR) describes
the m/f ratio of a mating pool.
The implicit sex ratios of the metoposaurs and
Eocyclotosaurus in this study are likely from all-adult, probably
water-bound, breeding populations (Lucas et al., 2010, 2016;
Rinehart and Lucas, 2016) (Table 1). In the Lamy and Rotten
Hill Koskinonodon samples, taphonomy studies have shown
that mass mortality events probably produced the fossils
under study (Lucas et al., 2010, 2016). Therefore, they do not
represent a single cohort and cannot be dened as an MSR.
There is no way to know if the representative animals have
bred at least once, which excludes their denition as an ESR.
If, as the above taphonomic and allometric studies indicate, all
adult metoposaurs joined a water-bound breeding population,
the dierence between ASR and OSR becomes moot. Although
taphonomic study of the Eocyclotosaurus bonebed at Tecolotito,
New Mexico showed a probable attritional assemblage rather
than mass mortality (Rinehart and Lucas, 2016), the implicit sex
ratio is the same—ASR or OSR, being essentially the same in
this circumstance.
Sex ratio: this study—The implicit sex ratios of the
metoposaurs vary from 1.2:1 to 2.7:1. In Koskinonodon, the best
represented metoposaur, the ratio is 1.9:1 and in Eocyclotosaurus,
1.3:1 (Table 1). But, is the bias male or female? We look to
574
FIGURE 10. A small sample of Eocyclotosaurus appetolatus ischia shows dimorphism in their width/length ratio (N = 6). A, A
histogram and a probability plot of w/l show bimodal behavior. Arrow marks the inection point between the two variant distributions.
B, NMMNH P- 67900, a left ischium in lateral view illustrated the low w/l (lunate) morph. NMMNH P-62767 shows the high w/l
(more triangular) morphology. This is a right ischium that has been horizontally ipped in Photoshop® for easier comparison.
575
FIGURE 11. Lateral skull angles of Sclerocephalus haeuseri show large variation, but do not show positively identiable dimorphism
in their lateral skull angles. A, A histogram and a probability plot show an approximately normal distribution of lateral skull angles
(N = 9). B, A scatterplot shows negative correlation of lateral skull angle to midline length. C, SMNS 58784 (upper), a medium
sized skull (81mm midline length) and SMNS 90055 (lower), the skull of the neotype of S. haeuseri representing a large skull (182
mm midline length) scaled to the same length illustrate probable ontogenetic shape change.
576
extant amphibians for guidance.
Extant amphibians—A literature search showed that
the sex ratios of extant amphibians and reptiles can be highly
variable between dierent populations of the same species, but
are normally male-biased (Lance et al., 2000; Alho et al., 2008)
(Table 2). With rare exceptions (e.g., the common frog, Rana
temporaria: Alho et al., 2008), anurans and the common toad,
Bufo bufo (Loman and Madsen, 2010), show strong male bias.
Salamanders, e.g., Salamandra atra atra, the alpine salamander,
and Ambystoma tigrinum, the tiger salamander, show male bias;
quite strong in A. tigrinum.
The giant salamanders (Cryptobranchidae) vary, depending
on population, from 1:1 up to 1.8:1 m/f ratio, but never show a
female bias in any of the studied populations (Hillis and Bellis,
1971; Peterson, 1988; Humphries and Pauley, 2005; Foster et
al., 2009; Browne et al., 2014) (Table 2).
For comparison, the archosaurian reptile, Alligator
mississippiensis, shows male bias (1.4:1) in a very large sample
of approximately 3000 animals (Lance et al., 2000) (Table 2).
Humans also show male bias, but to the much lesser extent of
1.04:1 (Jacobsen et al., 1999) (Table 1).
Bias—Jennions and Fromhage (2017) showed that the sex
ratios of diploid, sexually reproducing species deviate from 1:1
only under ecological pressures. So, some ecological factor(s)
produces almost exclusively male biased populations in modern
amphibians—especially in the cryptobranchids. If it is assumed
that wild amphibian populations of similar (salamander-like)
bauplan, and in the case of the giant cryptobranchids, nearly
similar size (large adult Andrias davidianus and A. japonicus
approach the size of a young adult metoposaur) encounter similar
ecological pressures, then the metoposaur and Eocyclotosaurus
study groups are probably male biased. The more common
dimorphs are plausibly male, and the rarer ones, female.
Hypothetical male and female morphs—Given the
presumption that the males are the more numerous gender, it
is now possible to postulate a partial suite of characters for the
presumed male and female morphs of Koskinonodon.
Males: 1. Comprise 65% of the population; 2. Have a high
TABLE 2. Sex ratios of some extant amphibians and the American alligator. Ratios for all extant amphibians are taken from
statistically signicant populations per references cited. The Alligator mississippiensis data are highly variable, 1.4:1 m/f is the
average of a very large database (N ≈ 3000).
Taxon Sex ratio m/f or f/m Reference
Bufo bufo (common toad) range 1.9:1
5.3:1
m/f Loman and Madsen, 2010
Rana temporaria (common frog) 2.3:1 f/m Alho et al., 2008
Salamandra atra atra (alpine
salamander)
1.5:1 m/f Romano et al., 2018
Ambystoma tigrinum (tiger salamander) 4.9:1 m/f Sever and Dineen, 1978
Andrias davidianus (Chinese giant
salamander)
1:1 = Luo et al., 2018
Cryptobranchus alleganiensis
(hellbender)
1.2:1 m/f Browne et al., 2014, citing: Humphries
and Pauley, 2005
Cryptobranchus alleganiensis
(hellbender)
1.6:1 m/f Browne et al., 2014, citing: Hillis and
Bellis, 1971
Cryptobranchus alleganiensis
(hellbender)
1:1 = Browne et al., 2014, citing: Peterson,
1988
Cryptobranchus alleganiensis
(hellbender)
1.8:1 m/f Browne et al., 2014, citing: Foster et
al., 2009
Alligator mississippiensis 1.4:1 m/f Lance et al., 2000
lateral skull angle resulting in a more acute snout and more
triangular skull (Fig. 3B); 3. Have a narrower clavicle (Fig.4C,
left); 4. Have a greater distal angle of the ilium (at least in
the Lamy population) (Fig.5C, lower); 5. Have a wider, more
triangular ischium (Fig. 6C, lower).
Females: 1. Comprise 35% of the population; 2. Have a
low lateral skull angle resulting in a broader snout and more
parallel lateral skull margins (Fig. 3A); 3. Have a wider clavicle
(Fig. 4C, right); 4. Have an ilium with a low distal angle (Lamy
population) (Fig. 5C, upper); 5. Possess ischia that are of the
narrower, more lunate morphology (Fig. 6C, upper).
Ethological implications—Parental care among the
salamanders, particularly among the giant cryptobranchids,
includes: defending the den in which the eggs are laid (dicult
to tell if it is the eggs, or the den itself that is being defended);
agitating the eggs, presumably to prevent yolk attachments; tail
fanning to increase the ow of oxygenated water over the eggs;
and eating of eggs that are dead, unfertilized, or infected with
mold, thereby preventing the spread of infection to healthy eggs,
termed “hygienic lial cannibalism” (Okada et al., 2015).
In the cryptobranchids, it is the males who perform these
duties. Females visit the den (typically a burrow in the riverbank),
which is occupied by a male “den-master,” lay their eggs, which
are externally fertilized, and leave. The male remains to occupy
the den, care for the eggs, and to receive other females (Okada
et al., 2015). In other genera of salamanders in which internal
fertilization is the norm, the females mate, and then lay their
eggs later. The male is not present at the egg laying, so it is
the female who remains to care for the eggs. All salamanders
except the cryptobranchids, hynobids, and sirenids have internal
fertilization by way of spermatophores (Okada et al., 2015,
citing Reinhard et al., 2013), and in these species the females
are the care givers (Jennions and Fromhage, 2017).
CONCLUSIONS
A statistical study of the lateral skull angles of three
metoposaur species and one capitosaur shows two distinct
skull morphologies for these temnospondyl taxa—one more
577
triangular with a more acute snout, the other having a broader
snout with more parallel sides (Figs. 3, 7C, 8C, 9C). In the
metoposaur Koskinonodon, we found two distinct morphologies
in some postcranial bones that occur in proportions similar
(within sample size error) to the two skull morphs.
Sexual dimorphism or mixed species could account for such
ndings, but we have presented arguments against the presence
of mixed species. The most parsimonious explanation of the two
morphologies is that they represent sexual dimorphs. We believe
that the evidence presented here makes a strong case for sexual
dimorphism in the Temnospondyli.
Using extant amphibians, especially the giant cryptobranchid
salamanders as analogs, it is most probable that the morph that
comprises the greater part of the dimorphic population represents
the males. The above-mentioned likelihood that the males have
the wider pelvis seems counterintuitive, but as amphibian eggs
are anamniotic and shell-less, it is unlikely that females would
need a wider pelvis for the purpose of egg laying.
Based on the work of Jennions and Fromhage (2017),
Okada, et al. (2017), and others, we speculate that if parental
care was present in the extinct study species, as it is in extant
amphibians, the more numerous males were probably the
parental care givers, and the rarer females were probably more
discriminating in mate selection.
ACKNOWLEDGMENTS
We thank diverse collections managers and curators for
access to specimens at their institutions. Andrew Heckert
photographed many of the MCZ and NMNH specimens used
in this study. Justin Spielmann photographed many of the
MNHN and YPM specimens used here. The authors thank
Andrew Heckert and Adrian Hunt for thoughtful reviews that
resulted in improvements to this manuscript. Life reconstruction
artwork of the study animals was executed by Mathew Celeskey
(Koskinonodon perfectum), Fredrik Spindler (Eocyclotosaurus
appetolatus), and Dmitry Bogdanov (Sclerocephalus haeuseri).
REFERENCES
Alho, J. S., Herczeg, G., and Merilä, J., 2008, Female-biased sex ratios
in subarctic common frogs: Journal of Zoology, v. 275, p. 57-63.
Antczak, M. and Bodzioch, A., 2018, Ornamentation of dermal bones
of Metoposaurus krasiejowensis and its ecological implications:
PeerJ 6:e5267; DOI 10.7717/peerj.5267.
Bonczarowska, J. H., McWhirter, Z., and Kranioti, E. F., 2021, Sexual
dimorphism of the lateral angle: Is it really applicable in forensic
sex estimation?: Archives of Oral Biology, v. 124, 105052.
Browne, R. K., Li, H., Wang, Z., Okada, S., Hime, P., McMillan, A.,
Wu, M., Diaz, R., McGinnity, D., and Briggler, J. T., 2014, The
giant salamanders (Cryptobranchidae): Part B. Biogeography,
ecology, and reproduction: Amphibian and Reptile Conservation,
v. 5, p. 30-50.
Brusatte, S. L., Butler, R. J., Mateus, O., and Steyer, J. S., 2015, A new
species of Metoposaurus from the Late Triassic of Portugal and
comments on the systematics and biogeography of metoposaurid
temnospondyls, Journal of Vertebrate Paleontology, DOI:
10.1080/02724634.2014.912988.
Colbert, E. H. and Imbrie, J., 1956, Triassic metoposaurid amphibians:
Bulletin of the American Museum of Natural History, v. 110, p.
405-452.
Clack, J. A., 2012, Gaining Ground; the Origin and Evolution of
Tetrapods. Indiana University Press, Bloomington, 523 p.
Damiani, R., 2008, A giant skull of the temnospondyl Xenotosuchus
africanus from the Middle Triassic of South Africa and its
ontogenetic implications: Acta Palaeontologica Polonica, v. 53,
p.75-84.
Dias, E. V., and Richter, M., 2002, On the squamation of Australerpeton
cosgri Barberna, a temnospondyl amphibian from the Upper
Permian of Brazil: Anais da Academia de Ciências, v. 73 p. 477-
490.
Dutuit, J.-M., 1976, Introduction à ľétude palaéontologique du Trias
continental Marocain. Description des premiers Stegoceohales
recueillis dans couloir ďArgana (Atlas Occidental) : Memoires du
Museum National D’Histoire Naturelle, 253 p.
Fairbairn, D. j., 1997, Allometry for sexual size dimorphism: Pattern
and process in the coevloution of body size in males and females:
Annual Review of Ecological Systems, v. 28, p. 659-687.
Fairbairn, D. J., Blankenhorn, W. U., and Székely, T., 2007, Sex, Size,
and Gender Roles: Evolutionary Studies of Sexual Dimorphism.
Oxford University Press, 266 p.
Ferrario, V. F., Sforza, C., Pizzini, G., Vogel, G., and Miani, A., 1993,
Sexual dimorphism in the human face assessed by Euclidean
distance matrix analysis: Journal of Anatomy, v. 183, p. 593-600.
Foster, R. L., McMillan, A. M., and Roblee, K. J., 2009, Population
status of Hellbender salamanders (Cryptobranchus alleganiensis)
in the Allegheny River drainage of New York State: Journal of
Herpetology, v. 43, p. 579-588.
Geary, D. C., 1998, Male, Female: The Evolution of Human Sex
Dierences. American Psychological Association, Washington,
D.C., 397 p.
Hillis, R. E., and Bellis, E. D., 1971, Some aspects of the ecology of
the hellbender, Cryptobranchus alleganiensis, in a Pennsylvania
stream: Journal of Herpetology, v. 5, p. 121-126.
Holden, C., and Mace, R., 1999, Sexual dimorphism in stature and
women’s work: a phylogenetic cross-cultural analysis: American
Journal of Physical Anthropology, v. 110, p. 27-45.
Humphries W. J., and Pauley, T. K., 2005, Life history of the hellbender,
Cryprobranchus alleganiensis, in a West Virginia stream: The
American Midland Naturalist, v. 154, p. 135-142.
Hunt, A. P., 1993, Revision of the Metoposauridae (Amphibia:
Temnospondyli) and description of a new genus from North
America: Museum of Northern Arizona, Bulletin 59, p. 67-97.
Huxley, J. S., 1966, Introduction to a discussion of ritualization of
behavior in animals and man: Philosophical Transactions of the
Royal Society of London B, v. 251, p. 249-271.
Jacobsen, R., Moller, H., and Mouritsen, A., 1999, Natural variation in
the human sex ratio: Human Reproduction, v. 14, p. 3120-3125.
Jennions, M. D., and Fromhage, L., 2017, Not all sex ratios are equal: the
Fisher condition, parental care and sexual selection: Philosophical
Transactions of the Royal Society B, v. 372, 20160312, http://
dx.doi.org/10.1098/rstb.2016.0312.
JMP 10®, 2012, Statistics and Data Analysis, Version 10.0.0, 2012,
SAS Institute Inc., Cary NC.
KaleidaGraph®, 2010, Graphing and Data Analysis, Version 4.1.1,
Synergy Software, Reading, PA.
King, J. R., 1971, Probability Plots for Decision Making: Industrial
Press Inc., New York, 290 p.
Kirchengast, S., 2014, Human sexual dimorphism – a sex and gender
perspective: Anthropologische Anzeiger, v. 71, p. 123-133.
Kozerska, M., Skrzat, J., and Szczepanek, A., 2015, Application of the
temporal bone for sex determination from the skeletal remains:
Folia Medica Cracoviensia, v. 55, p. 33-39.
Lance, V. A., Elsey, R. M., and Lang, J. W., 2000, Sex ratios of American
alligators (Crocodylidae): male or female biased?: Journal of the
Zoological Society of London, v. 252, p. 71-78.
Loman, J., and Madsen, T. R., 2010, Sex ratio of breeding common
toads (Bufo bufo) – inuence of survival and skipped breeding:
Amphibia-Reptilia, v. 31, p. 509-524.
Lucas, S. G., Rinehart, L. F., Krainer, K., Spielmann, J. A., and Heckert,
A. B., 2010, Taphonomy of the Lamy amphibian quarry: A Late
Triassic bonebed in New Mexico, U.S.A.: Paleogeography,
Paleoclimatology, Paleoecology, v. 289, p. 388-398.
Lucas, S. G., Rinehart, L. F., Heckert, A. B., and Spielmann, J. A., 2016,
Rotten Hill: A Late Triassic bonebed in the Texas Panhandle, USA:
New Mexico Museum of Natural History and Science, Bulletin
71, 97 p.
Luo, Q., Tong, F., Song, Y., Wang, H., Du, M., and Ji, H., 2018,
578
Observation of the breeding behavior of the Chinese giant
salamander (Andrias davidianus) using a digital monitoring
system: Animals, www.mdpi.com/journal/animals, doi:10.3390/
ani8100161, v. 8, 13 p.
Mallon, J. C., 2017, Recognizing sexual dimorphism in the fossil
record: lessons from nonavian dinosaurs: Paleobiology, v. 43,
p.495-507.
Okada, S., Fukuda, Y., and Takahashi, M. K., 2015, Paternal care
behaviors of Japanese giant salamander Andrias japonicus in
natural populations: Journal of Ethology, v. 33, p.1-7.
Olson, E. C., 1951, Diplocaulus, a study in growth and variation:
Fieldiana, Geology, v. 11, p. 57-154.
Petaros, A., Gavin, H. M., Sholts, S. B., Schlager, S., and Wärmländer,
S.K. T. S., 2017, Sexual dimorphism and regional variation in
human frontal bone inclination measured via 3D models: Legal
Medicine, v. 29, p. 53-61.
Peterson, C. L., 1988, Breeding activities of the hellbender in Missouri:
Herpetological Review, v. 19, p. 28-29.
Reinhard, S., Voitel, S., and Kupfer, A., 2013, External fertilisation and
paternal care in the paedomorphic salamander Siren intermedia
Barnes, 1826 (Urodela: Sirenidae): Zoologischer Anzeiger, v. 235,
p. 1-5.
Rinehart, L. F., and Lucas. S. G., 2001, A statistical analysis of a
growth series of the Permian nectridean Diplocaulus magnicornus
showing two-stage ontogeny: Journal of Vertebrate Paleontology,
v. 21, p. 803-806.
Rinehart, L. F., and Lucas, S. G., 2016, Eocyclotosaurus appetolatus,
a Middle Triassic amphibian: osteology, life history, and
paleobiology: New Mexico Museum of Natural History and
Science, Bulletin 70, 118 p.
Rinehart, L. F., Lucas, S. G., and Schoch, R. R., 2015, Eocyclotosaurus
appetolatus, a new cyclotosaurid amphibian from the Middle
Triassic (Perovkan) Moenkopi Formation of New Mexico, U.S.A.:
Journal of Vertebrate Paleontology, 15 p. http://dx.doi.org/10.1080
/02724634.2014.929140.
Rinehart, L. F., Lucas, S. G., Heckert, A. B., Spielmann, J. A., and
Celeskey, M. D., 2009, The paleobiology of Coelophysis bauri
(Cope) from the Upper Triassic (Apachean) Whitaker quarry,
New Mexico, with detailed analysis of a single quarry block: New
Mexico Museum of Natural History and Science, Bulletin 45, 260
p.
Romano, A., Anderle, M., Forti, A., and Partel, P., 2018, Population
density, sex ratio and body size in a population of Salamandra
atra atra on the Dolomites: Acta Herpetologica, v. 13, p. 195-199.
Romer, A. S., 1939, An amphibian graveyard: Scientic Monthly, v.
49, p. 337-339.
Sanchez, S., and Schoch, R. R., 2013, Bone histology reveals a high
environmental and metabolic plasticity as a successful evolutionary
strategy in a long-lived homeostatic Triassic temnospondyl,
Evolutionary Biology, v. 40, p. 627-647.
Sever, D. M., and Dineen, C. F., 1978 Reproductive ecology of the
tiger salamander, Ambystoma tigrinum, in northern Indiana:
Proceedings of the Indiana Academy of Science, v. 87 p. 189-203.
Schoch, R. R. and Milner, A. R., 2000, Stereospondyli: Handbook of
Paleoherpetology, Part 3B, Verlag Dr. Friedrich Pfeil, Munich, 203
p.
Schoch, R. R., 2009, Developmental evolution as a response to diverse
lake habitats in Paleozoic amphibians: Evolution, v. 63, p. 2738-
2749.
Shoch R. R., and Witzmann, F., 2009, Osteology and relationships of
the temnospondyl genus Sclerocephalus: Zoological Journal of the
Linnean Society, v. 157, p. 135-168.
Schoch, R. R., 2014, Amphibian Evolution: the Life of Early Land
Vertebrates. John Wiley and Sons, West Sussex, U.K., 264 p.
Simpson, G. G., Roe, A., and Lewontin, R. C., 1960, Quantitative
Zoology, Harcourt, Brace & World, Inc., New York, 440 p.
Sinclair, A. J., 1976, Applications of Probability Graphs in Mineral
Exploration: Special Volume No. 4, The Association of Exploration
Geochemists, Richmond Printers Ltd., Richmond, B.C., Canada,
95 p.
Sulej, T., 2002, Species discrimination of the Late Triassic temnospondyl
amphibian Metoposaurus diagnosticus: Acta Palaeontologica
Polonica, v. 47, p. 535-546.
Sulej, T., 2007, Osteology, variability, and evolution of Metoposaurus,
a temnospondyl from the Late Triassic of Poland: Palaeontologia
Polonica, v. 64, p. 29-139.
Teschner, E. M., Sander, P. M., and Konietzko-Meier, D., 2017,
Variability of growth pattern in Metoposaurus krasiejowensis
humeri and its biological meaning: Journal of Iberian Geology v.
44, p. 99-111.
Valen, L. Van, 2005, The statistics of variation; in Hallgrímsson, B.
and Hall, B. K., eds., Variation: a Central Concept in Biology,
Elservier Academic Press, Amsterdam, 568 p.
Wahl, J., 1981, Ein Beitrag zur metrischen Geschlechtsdiagnose
ver brannter und unverbrannter menschlicher Knochenreste-
ausgearbei tet an der Pars petrosa ossis temporalis: Zeitschrift
Rechtsmedizin, v. 86, p. 79–101.
Werneburg, R., Schneider, J., and Lucas S. G., 2013, The dissorophoid
Milnererpeton huberi (Temnospondyli) from the :Late
Pennsylvanian Kinney Brick Quarry in New Mexico restudied –
paleontology, paleoenvironment, and age: New Mexico Museum
of Natural History and Science, Bulletin 59, p. 349-369.
Zeigler, K. E., Lucas, S. G. and Heckert, A. B., 2002, The Late Triassic
Canjilon quarry (Upper Chinle Group, New Mexico) phytosaur
skulls: Evidence of sexual dimorphism in phytosaurs: New Mexico
Museum of Natural History and Science, Bulletin 21, p. 179-188.
Zeigler, K. E., Heckert, A. B., and Lucas, S. G., 2003, Phytosaur
(Archosauria: Parasuchidae) cranial and mandibular material from
the Upper Triassic Snyder Quarry (Petried Forest Formation,
Chinle Group): New Mexico Museum of Natural History and
Science, Bulletin 24, p. 81-88.
Zittel, K. Von, 1887-1890, Handbuch der Paläozoologie, 3,Vertebrata
(Pisces, Amphibia, Reptilia, Avies): Oldenbourg, München und
Leipzig, 1890 p.
579
APPENDIX 1
Koskinonodon perfectum skull metrics from the Lamy and Rotten Hill populations (N = 26). Midline length in mm. Lateral
skull angle in degrees.
Locality Institution Specimen Midline
length
Lateral skull
angle
Lamy NMMNH P-69257 472 27
Lamy NMMNH P-69254 426 35
Lamy NMMNH P-69251 382 34
Lamy NMMNH P-66748 430 33
Lamy NMMNH P-67334 379 32
Lamy NMNH 6 294 25
Lamy NMNH 3 364 34
Lamy NMNH 2 452 31
Lamy MCZ 1674-8 502 24
Lamy MCZ 1674-12 426 29
Lamy MCZ 1674-10 25
Lamy MCZ 1674-3 30
Lamy MCZ 1674-6 398 24
Lamy MCZ 1674-4 388 30
Lamy NMMNH P-61420 522 24
Lamy NMMNH P-67335 472 40
Rotten Hill PPHM WT-3166-1 392 26
Rotten Hill PPHM 369 26
Rotten Hill PPHM PPHM 6 432 30
Rotten Hill PPHM PPHM 5 505 28
Rotten Hill PPHM 367 26
Rotten Hill PPHM WT 3067-2 35
Rotten Hill PPHM WT 3067 430 31
Rotten Hill PPHM 32
Rotten Hill PPHM 3114 365 34
Rotten Hill PPHM WT 3011 367 30
580
APPENDIX 2
Koskinonodon perfectum clavicle metrics from the Lamy amphibian quarry (N = 14). Metrics in mm.
Institution Specimen Length Width W/L
NMMNH P-62626 442 230 0.520362
NMMNH P-69393 370 170 0.459459
NMMNH P-69636 355 160 0.450704
NMMNH P-64645 345 163 0.472464
NMMNH P-69685 364 175 0.480769
NMMNH P-69724 410 180 0.439024
NMMNH P-69374 340 159 0.467647
NMMNH P-69343 374 180 0.481283
NMMNH P-69589 315 135 0.428571
NMMNH P-64381 343 154 0.44898
MCZ C 415 184 0.443373
MCZ L 263 135 0.513308
MCZ N 290 160 0.551724
YPM 390 195 0.5
APPENDIX 3
Koskinonodon perfectum ilium metrics from the Lamy population (N = 24). Metrics in mm. Angles in degrees.
Institution Specimen Anterior length Base a-p width Distal angle
MCZ 1925-1 74 36 11
MCZ 1925-2 79 44 10
MCZ 1925-3 79 40 5
MCZ 1925-4 94 48 7
MCZ 1925-5 99 47 8.5
MCZ 1925-6 94 47 0
MCZ 1925-7 76 38 15
MCZ 1925-8 79 37.5 4
MCZ 1925-9 73 32 7
MCZ 1925-10 72 34 15
MCZ 1925-11 65 35 5
MCZ 1925-12 80 44 20
MCZ 1925-13 86 45 15.5
MCZ 1925-14 97 47 14
MCZ 1925-15 98 43 12
MCZ 1925-16 95 49 5.5
MCZ 1925-17 105 61 17.5
MCZ 1925-18 80 42 5.5
MCZ 1925-19 76 39 17
MCZ 1925-20 74 40 27
MCZ 1925-21 71 34 8
NMMNH P-58204 72 42 14
NMMNH P-68093 105 54 8
NMMNH P-58306 73 40.5 12
581
APPENDIX 4
Koskinonodon perfectum length/width ratio of the ischium from the Lamy population (Rotten Hill population below) (N = 16).
Metrics in mm.
Institution Specimen Length Width W/L
MCZ 2013-1 72 52 0.722222
MCZ 2013-2 60 53 0.883333
MCZ 2013-3 69 61 0.884058
MCZ 2013-4 78.5 69 0.878981
MCZ 2013-5 49 36 0.734694
MCZ 2013-6 62 54 0.870968
NMMNH P-63480 80 67 0.8375
NMMNH P-67380 75 59 0.786667
NMMNH P-67533 69 50 0.724638
NMMNH P-58303 64 52.5 0.820313
NMMNH P-67532 69.00 50.00 0.724638
NMMNH P-60224 79.3 55.7 0.702396
NMMNH P-60118 71.6 57 0.796089
Rotten Hill population
PPHM 3138a 55 51 0.927273
PPHM 3096 59 52 0.881356
PPHM 3138b 45 42 0.933333
APPENDIX 5
Dutuitosaurus ouazzoui lateral skull angles and total lengths (includes exoccipitals) from Dutuit (1976). Lengths in mm, angles
in degrees.
Institution Specimen Total
length
Lateral
skull angle
MNHN XIII/13/66 (1) 345 29.1
MNHN XIII/36/66 28.7
MNHN XIII/18/66 28.1
MNHN XIII/16/66 29
MNHN XIII/14/66 30.6
MNHN XIII/14/66 (4) 425 30.2
MNHN XIII/14/66 (5) 315 30.8
MNHN XIII/14/66 (6) 480 30.6
MNHN XIII/4/66 30.1
MNHN XIII/12/65 (1) 670 27.5
MNHN XIII/38/65 L 32.4
MNHN XIII/38/65 S 28.2
MNHN XIp/4/66 28.5
582
APPENDIX 6
Metoposaurus diagnosticus lateral skull angles measured from Sulej (2007). Midline lengths from Sulej (2007; appendix 1).
Lengths in mm, angles in degrees.
ZPAL Specimen Midline length
Lateral
skull
angle
AbIII/1683 276 28.9
AbIII/682 278 32.2
AbIII/1514 32.4
AbIII/1682 290 29.3
AbIII/894 291 30.9
AbIII/1191 318 29.7
AbIII/1674 320 32
AbIII/681 322 31.6
AbIII/358 323 32.9
AbIII/3 375 34.1
AbIII/1192 470 33
APPENDIX 7
Eocyclotosaurus appetolatus from the Tecolotito, New Mexico bonebed. Tabulated midline lengths and lateral skull angles.
Midline length in mm. Lateral skull angle in degrees.
Specimen Midline length Lateral
skull angle
uncat J-2 295 28.7
uncat J-2 290 30
P-64166 389 26.4
P-64126 31.5
P-66832 472 28.7
P-54360 28.3
P-63328 227 28.2
P-66861 241 32.8
P-64408 33.1
P-65262 318 33.8
P-68626 31.8
P-69165 28.4
P-67401 310 35.7
in prep 27.9
APPENDIX 8
Eocyclotosaurus appetolatus ischium width/length ratios (N = 6).
Institution Specimen W/L
NMMNH P-68081 0.84
NMMNH P-67426 0.89
NMMNH P-62679 0.81
NMMNH P-67900 0.7
NMMNH P-61422 0.72
NMMNH P-62767 0.85
583
APPENDIX 9
Lateral skull angles of selected Sclerocephalus hauseri specimens (N = 9). Lengths in mm. Angles in degrees.
Institution Specimen Midline
length
Lateral
skull
angle
UGKU Odenheim/Glan 32 44.6
UGKU Odenheim/Glan 30 50.2
UGKU Niederhausen 110 38
SMNS Neotype 90055 182 28.9
SMNS 58784 81 38.9
SMNS 58989 15 37
SMNS 91283 149 41.2
SMNS 91280 190 18
SMNS 91282 97 43.9
