Acta Theriologica 44 (3): 309-320, 1999.
PL ISSN 0001-7051
Ulnar dimensions and fossoriality in armadillos
Sergio F. VIZCAÍNO, Richard A. FARIÑA* and Gerardo V. MAZZETTA
Vizcaíno S. F., Fariña R. A. and Mazzetta G. V. 1999. Ulnar dimensions and fossoriality
in armadillos. Acta Theriologica 44: 309-320.
Ulnar dimensions were measured in 14 species of armadillos (Xenarthra: Dasy-
podidae). An index of fossorial ability (IFA) was constructed, relating the length of the
olecranon process to the remaining length of the ulna. For comparative purposes, the
same measurements were taken in 14 other species of mostly South American mammals
belonging to 3 orders and 11 families. The fossorial habits of these mammals were
classified into 3 categories: (1) species mostly cursorial and non-digging; (2) species that
often dig, but to which digging plays no essential part in their alimentary strategy and
are not burrowers; and (3) species that are burrowers. IFA means of the studied
mammal orders were compared using one-way analysis of variance on log-transformed
data. Bivariate size allometry between ulnar dimensions and body mass was assessed by
fitting (least squares and geometric mean) linear regressions of log-transformed data.
It is concluded that the IFA discriminates among the species according to their fossorial
habits within orders, but it is not equally useful in distinguishing fossorial species
between orders. In armadillos, the relationships between ulnar dimensions and body
mass are isometrical. Finally, the IFA is independent of body size.
CONICET, Departamento Científico Paleontología de Vertebrados, Museo de La Plata,
Paseo del Bosque s/n, 1900 La Plata, Argentina, e-mail: vizcaino@museo.fcnym.unlp.edu.ar
(SFV); Departamento de Paleontología, Facultad de Ciencias, Iguá 4225, 11400 Monte-
video, Uruguay, e-mail: fari~a@fcien.edu.uy; mazzetta@fcien.edu.uy (RAF, GVM)
Key words: armadillos, Dasypodidae, fossoriality, ulna length, olecranon length, body size
Introduction
Armadillos are armoured mammals classified in the family Dasypodidae
(Mammalia: Xenarthra). They are endemic of South America, and are widespread
on the sub-continent, with some 20 living species (Wetzel 1985). One of them,
Dasypus novemcinctus Linnaeus, 1758, reaches the southern USA. The group has a
long history, their remains being known from the middle Palaeocene of South
America (Patterson and Pascual 1972). Almost all species are, to some extent,
fossorial. Digging is accomplished chiefly by a strong extension of their forearms.
This paper will test a hypothesis of the relationship between the fossorial habits
of armadillos and their ulnar morphology. The relative length of the olecranon
seemed to be potentially revealing, for it has a well-known association with
fossorial habits (Shimer 1903). On the other hand, a short olecranon implies a
* To whom correspondence should be sent.
[309]
310
S. F. Vizcaíno et al.
selection for fast, rather than powerful, extension of the forearm (Fariña and
Blanco 1996). Further, that feature is easily studied both in living and fossil
species. Quantitative approaches have been developed in the past (Goldstein 1972,
Hildebrand and Hildebrand 1994). However, they have not been extensively
applied to South American mammals, nor has the problem of possible allometric
effects been properly addressed. The family Dasypodidae is an appropriate group
with which to test this hypothesis, because it is clearly monophyletic with species in
a variety of sizes.
For comparative purposes, other mammals were considered in this study,
involving other 14 species of mammals (mostly from South America) belonging to 3
orders and 11 families. Their habits range from cursorial to burrowing.
Material and methods
A total of 135 ulnae of several species belonging to 4 orders of (mostly) South American mammals
were used (Table 1). In the case of armadillos (77 individuals belonging to 14 species), they represent
the breadth of systematic diversity of the group at the tribe or subfamily level, following the
classification proposed by Scillato-Yané (1980). Also, these animals range through three orders of
magnitude in body mass from the 120 g pygmy armadillo Chlamyphorus truncatus Harlan, 1825 to the
almost 50 kg giant armadillo Priodontes maximus (Kerr, 1792). The species used in this study were
housed in the collections of the institutions listed in Appendix 1.
The fossorial habits of the animals studied were classified into 3 categories, namely mostly
cursorial (sensu Jenkins 1971 and Stein and Casinos 1997: "cursorial mammals are those terrestrial
quadrupeds that posses vertically-oriented limbs which move in a parasagittal plane, regardless of the
gait being employed") and non-digging mammals (category 1); species that often dig, but to which
digging plays no essential part in their alimentary strategy and are not burrowers (category 2); and
species that are burrowers or that feed on termites or ants (category 3), as in both cases the digging
movements of the forearm are the same, namely, a strong extension of the forearm. The choice for
including a particular species in one of those three categories was accomplished according to data
gathered from the literature (for instance, Nowak 1991) and from personal field observations. The
naked-tailed armadillos Cabassous spp. and the pygmy Chlamyphorus truncatus have extremely
fossorial habits (Nowak 1991). The giant armadillo Priodontes maximus is considered a powerful and
rapid digger, and shelters in burrows of its own construction (Nowak 1991).
The three-banded armadillo Tolypeutes matacus (Desmarest, 1804) is the most cursorial within the
family. Nowak (1991) states that this species does not seem to dig holes.
Those armadillos of the category 2 belong to the subfamilies Dasypodinae and Euphrachtinae. All
these species for which habits are known have the average fossorial habits expected for the members of
the group. However, this can only be tentatively said about Dasypus kappleri Krauss, 1862, because
the habits of this species are not well known.
In Carnivora, the maned wolf Chrysocyon brachyurus (Illiger, 1815) and Geoffroy's cat Felis
geoffroyi d'Orbigny and Gervais, 1842 are non-diggers. The grey fox Dusicyon gymnocercus (Fischer,
1814) has been reported to have its dens in burrows made by other animals, such as viscachas and
armadillos (Nowak 1991). Finally, the skunks Conepatus spp. have the highest ratio and dig their own
burrow as well as using burrows dug by other animals (Redford and Eisenberg 1989).
While the huemul Hippocamelus bisulcus (Molina, 1782) like other deer is completely cursorial, the
wild boar Sus scrofa Linnaeus, 1758 has also the well-known digging habits of suids. To some extent,
this is valid for the peccary Tayassu pécari (Link, 1795).
Within the five species of rodents under analysis, the Patagonian hare Dolichotis patagonum
(Zimmerman, 1780) and the agouti Dasyprocta punctata Gray, 1842 are cursorial mammals (Nowak
1991). The intermediate behaviour is represented by the coypo Myocastor coypus (Molina, 1782). For
Fossoriality in armadillos
311
Table 1. Untransformed measurements of bones, index of fossorial ability (IFA, as explained in the
text), body mass and habits (as discussed in text) for some mammals. When appropriate, values are
given as mean ± SD.
Taxon
Ulnar length Olecranon length
IFA Body mass
Habits
Taxon
(mm) (mm)
(n)
(kg)
Tolypeutes matacus
49.4 ± 4.9
17.0
+
1.5
0.53 ± 0.04 (4) 1.53
1
Dasypus hybridus 52.9 ± 2.7 20.6
1.2 0.64 ± 0.03 (6)
2.04 2
Dasypus novemcinctus
68.7 ± 3.5 26.7
2.1 0.64 ± 0.06 (17) 3.30
2
Dasypus happier i
93.2 ± 2.9
36.1
+
2.0
0.64 ± 0.05 (4) 10.60
?
Euphrachtus sexcinctus 70.3 ± 4.3 27.9
+
4.0 0.66 ± 0.07 (14) 8.19
2
Chaetoph'actus vellerosus 42.7 16.5 0.63 (1) 1.10
2
Chaetoph~actus villosus 63.4 ±
2.8
25.2
+
2.7 0.68 ± 0.16 (5) 4.50
2
Zaedyus jichiy 44.4 ± 0.21 16.1 0.2 0.57 ± 0.01 (2)
1.74 2
Cabassou; chacoensis 48.6 27.1 1.26 (1) 1.55 3
Cabassou; centralis
55.4 ±
4.6 26.6 3.6
0.92 ± 0.14 (3) 3.80 3
Cabassou; unicinctus
57.6 ±
8.4
27.9 2.8
0.96 ± 0.12 (7)
3.50
3
Cabassou; tatouay 68.7 31.4 0.84 (1) 6.20 3
Priodontes maximus 132.6 ± 4.0 62.8 6.8 0.91 ± 0.15 (10) 45.19 3
Chlamypr.orus truncatus
22.2 ±
0.1 11.9 0.1 1.15 ± 0.04 (2)
0.12 3
Conepatui chinga 62.0 13.5 0.28 (1)
-
2
Conepatu; sp. 60.1 ± 6.7 12.8 2.0 0.27 ± 0.03 (7)
-
2
Felis geofroyi 110.8 ± 23.5 14.3
±
1.6 0.15 ± 0.03 (4) -
1
Chrysocycn brachyurus 320.4 ±
50.0
28.7
3.1
0.10 ± 0.01 (9)
-
1
Dusicyon gymnocercus 136.4 ±
2.3
17.1
•±
1.0
0.14 ± 0.01 (8)
-
2
Hippocanelus bisulcus
270.0 55.0 0.26 (1)
-
1
Sus scrofc,
185.0 70.0 0.61 (1) -
2
Tayassu jecari 128.0
36.0 0.39 (1)
-
2
Dasyproca sp. 72.5
11.5
0.19 (1)
-
1
Dasyproca punctata
80.7 ±
4.2
14.0
±
0.8 0.21 ± 0.02 (6)
-
1
Dolichoti: patagonum
170.0 ± 5.3 25.0
±
2.1
0.17 ± 0.01 (6)
-
1
Myocasto: coypus 91.3 ± 15.2 18.9
±
4.4 0.26 ± 0.03 (2)
-
2
Lagostomus maximus
86.6 ± 7.4 18.2
±
4.3 0.27 ± 0.06 (7) - 3
Hydrochceris hydrochaeris 166.8 ±
7.8 46.3
±
2.5 0.39 ± 0.01 (4)
-
1
shelter, tiis rodent takes over the hole of another animal or constructs its own burrow. The latter may
be a singe tunnel or a complex system containing passages that extend 15 m or more and chambers
(Nowak 1991). Viscacha Lagostomus maximus (Desmarest, 1817) constructs extensive burrow systems,
called "vEcacheras", some of which are used for centuries and may cover up to 600 m . To construct
these burows some 8 m
3
have to be moved, which illustrates the digging ability of this species (Nowak
1991). Fnally, the capybara Hydrochaeris hydrochaeris (Linnaeus, 1766) is not known to dig
whatsoever.
The neasurements taken on this material are shown in Fig. 1. Total ulnar length (UL) was
measurec from the tip of the olecranon to the tip of the styloid process, and olecranon length (OL) as
the distaice from the tip of the olecranon to a point the middle of the semilunar notch. Ulnar and
olecranoi lengths were measured in cleaned bones of adult individuals with appropriate callipers to
the neartst 0.1 mm. Also, lengths were measured on the right forearm whenever the landmarks were
intact. If they were not, the left forearm was used instead.
312
S. F. Vizcaíno et al.
The index of fossorial ability (IFA) was obtained by dividing
OL by the subtraction UL minus OL. This index describes the
relative mechanical advantage of the triceps muscle (Hildebrand
and Hildebrand 1994). This muscle is the main extensor of the
forearm, and is extensively used by those mammals that dig
with movements of the whole arm.
Univariate statistics used are described in standard textbooks
(Sokal and Rohlf 1981) and listed for each species in Table 1.
IFA means of the studied mammal orders were compared using
one-way analysis of variance (ANOVA) on log-transformed data
and Scheffe's multiple-range test at the 0.05 level of significance.
The same analyses were computed among subfamilies of Dasy-
podidae and also among families in each of the remaining
orders. Data were log-transformed to stabilize the variance.
The Scheffe's test was used because it is conservative, requiring
larger differences between population means for significance
ments of total ulnar length (UL, a)
thjm mQst mu
l
t
ipl
e
-comparison methods.
obtlined^
11011
^^ ^
&>
^ Bivariate size allometry was assessed by fitting the following
allometric model for two random variables X and Y (Huxley
1932, Gould 1966, 1971, notation follows Ricker 1984): Y= aJC.
This power equation was linearised by log-transformation
(base 10) into log Y = loga + b logX. Considerable controversy
surrounds the statistical methods used to estimate the two parameters, log a (intercept) and b (slope or
allometric coefficient), although most applications have been based on simple (least squares) linear
regressions of log Fon logX This method is easy to use and permits calculation of standard error (SE)
of the slope estimate but assumes, usually wrongly, that only Y is subject to measurement error or
natural variability or both. Geometric mean regressions (GMR) provide a slope which is the geometric
mean of the estimate derived from log Y (linearly) regressed on log X and the inverse of that derived
from regressing log X on log Y (Ricker 1973, 1984). Both methods were employed (see Swartz and
Biewener 1992) in order to test the relationship of forearm skeletal dimensions and body size in
armadillos. Hence, the following regressions of log-transformed data were calculated: ulnar and
olecranon length versus body mass, and olecranon length versus ulnar length (also performed for
comparison in each of the remaining mammal orders considered). Each slope (using the 95 %
confidence limits) was tested against the null hypothesis of isometry, ie the maintenance of geometric
similarity with size increase: [3 = 1/3 for osteometric lengths against body mass, and ¡i = 1 for
olecranon length as compared to ulnar length. Also, the relationship of IFA and body size in armadillos
was analysed.
Body masses were not available for all the studied individuals. Nevertheless, the masses of most
species belonging to the family Dasypodidae were obtained and averaged from the literature (Rood
1970, McNab 1980, Wetzel 1985, Fariña and Vizcaino 1997). The mass of Cabassous chacoensis Wetzel,
1980 was estimated by scaling down that of C. tatouay (Desmarest, 1804) (see Redford and Eisenberg
1989).
The bivariate relationships between IFA and the fossorial categories (considering all the species
studied or the species of each order separately) were explored with least-squares linear regressions of
log-transformed data.
We use nomenclature as in general books on mammals, such as Nowak (1991).
Fig. 1. Way in which the measure-
Results
ANOVA and Scheffe's multiple-range test showed that IFA means differ
significantly among all orders studied (F = 232.65, df = 3, 131, p < 0.0001).
Moreover, the variability among orders was greater than the variability within each
Fossoriality in armadillos 313
order, as evidenced by their relative sums of squares (59.7 and 11.2, respectively).
IFA values of the Dasypodidae were the highest, with a mean (± SD) of 0.74 ± 0.18
(range 0.47-1.26, n = 77). Among them, the three-banded armadillo Tolypeutes
matacus is in the lowest extreme, while the naked-tailed Cabassous chacoensis and
the pigmy armadillo Chlamyphorus truncatus are the species whose values are the
highest. The rest of the dasypodids yielded intermediate figures.
The species belonging to order Carnivora shows the lowest IFA values, with a
mean of 0.17 ± 0.07 (range 0.08-0.32, n = 29). The skunks Conepatus spp. have the
highest values, whereas the index for the maned wolf Chrysocyon brachyurus is the
lowest. The grey fox Dusicyon gymnocercus and the Geoffroy's cat Felis geoffroyi lie
between them, and closer to the latter.
The studied species of Rodentia have higher IFA values than those observed in
Carnivora, with a mean of
0.25
± 0.08 (range 0.16-0.39, n = 26). The rodents range
from the Patagonian hare Dolichotispatagonum and the agoutis Dasyprocta spp. in
the lower extreme to the capybara Hydrochaeris hydrochaeris at the upper limit.
The coypo Myocastor coypus and the viscacha Lagostomus maximus are at an
intermediate level.
IFA values of the studied species of Artiodactyla were higher than those of
Rodentia, with a mean of 0.42 ± 0.18 (range 0.26-0.61, n = 3). Among them, the
huemul Hippocamelus bisulcus has the lowest value, the wild boar Sus scrofa, the
highest and the peccary Tayassu pécari has an intermediate value.
ANOVA and Scheffe's test performed on Dasypodidae showed that IFA means of
the subfamilies were significantly different (F = 44.61, df = 4, 72, p < 0.0001),
except for Dasypodinae and Euphrachtinae. In Carnivora, IFA means differ
significantly among all families studied (F = 59.75, df = 2, 26, p < 0.0001). Similar
results were observed in Rodentia (F = 24.16, df = 4, 21, p < 0.0001), except for
Myocastoridae and Chinchillidae. It was not possible to compare IFA means of the
three families of Artiodactyla due to the very low number of specimens (n = 1, each
family).
In armadillos, the size-allometric relationship between ulnar length and body
mass exhibits isometry (b = 0.30, SE = 0.02,p > 0.10, r = 0.98, n = 14, Fig. 2a). The
same was observed for olecranon length versus body mass (b = 0.29, SE = 0.04, p >
0.25, r = 0.92, n = 14, Fig. 2b). Also, the relationship between olecranon length and
ulnar length shows no change of shape with size increase {b = 0.94, SE = 0.10,p >
0.50, r = 0.93, n = 14). The same is valid for the species of Rodentia (6 = 1.16, SE =
0.30, p > 0.60, r = 0.89, n = 6). In Carnivora a strong negative allometry was
observed between these forearm dimensions (6 = 0.46, SE = 0.08, p < 0.01, r =
0.96, n = 5). For the species of Artiodactyla, olecranon length was not significantly
correlated with ulnar length (r = 0.62, n = 3, p > 0.55), although the very small
sample could obscure the relationship. Considering all the species studied, the slope
of the regression between olecranon length and ulnar length is significantly
different from isometry (b = 0.50, SE = 0.13,p < 0.001, r = 0.59, n = 28, Fig. 3). In
all the analysed size-allometric relationships, but in the last two cases, very similar
