Testudo graeca tripolitania, a new taxon of spur-thighed tortoise found in Libya at risk of exploitation for the international pet trade

Abstract

The spur-thighed tortoise, Testudo graeca, is a widely distributed and deeply diversified species inhabiting areas of Africa, Asia, and Europe. For decades, populations of T. graeca from North Africa have been exploited for the international pet market. In particular, T. graeca of Libyan origin have been commercially imported into the United States several times since 2021. Using mitochondrial DNA (mtDNA) sequencing, we show that these imported tortoises include T. g. cyrenaica and a novel lineage originally reported in 2017 from a displaced specimen from a market in Libya. That novel lineage inhabits northwestern Libya, and we show that wild populations near Gharyan share the same mtDNA haplotype as those now sold as pets in the USA. Populations of T. graeca in northwestern Libya, previously identified as T. g. nabeulensis, are reassigned to a new subspecies Testudo graeca tripolitania described herein. Additional field work is needed to determine the mtDNA haplotypes of populations in central and southern Tunisia and to locate the subspecies boundary between T. g. nabeulensis and T. g. tripolitania.

Full text

Basic and Applied Herpetology 00 (2024) 000-000
DOI: http://dx.doi.org/10.11160/bah.279 Supplementary material available online
The spur‐thighed tortoise, Testudo grae-
ca, Linneaus, 1758, is a diverse species
with a large native range, extending ap‐
proximately 6700 kilometers from Moroc‐
co to Iran, and inhabits a wide range of
climatic conditions (TÜrkozan et al., 2023).
The divergence of T. graeca is generally
acknowledged to have occurred on multi‐
ple temporal scales as the species’ range
expanded over time (Fritz et al., 2009,
Graciá et al., 2017a). The differentiation of
six primary clades (five in Asia, and the
other one in North Africa and Europe)
occurred first during the Pliocene (7.95‐
3.48 million years ago). The Asian clades
are currently recognized as T. g. armeniaca,
T. g. buxtoni, T. g. ibera, T. g. terrestris, and
T. g. zarudnyi. Differentiation within the
North African clade occurred later, within
the early‐ to mid‐Pleistocene (3.47‐1.44
million years ago), to yield five North Afri‐
can subspecies: T. g. graeca, T. g. marroken-
sis, T. g. whitei, T. g. nabeulensis, and T. g.
cyrenaica. The nomenclature of T. graeca
was disrupted in 2020, with the discovery
that the type location of T. graeca Linneaus
Testudo graeca tripolitania,
a new taxon of spur-thighed
tortoise found in Libya at risk of exploitation for the
international pet trade
Stephen F. Poterala1,*, Paul Rattay1, Aaron S. Johnson1, Murad S.A. Buijlayyil2, Askin Kiraz2,
Ahmad M.S. Ajaj3
1 Turtle and Tortoise Preservation Group. 1042 N Higley Rd. Suite. 105, Mesa, AZ, USA.
2 Near East University. Near East Boulevard, Nicosia 99138, Cyprus.
3 Albarari Organization for Conservation of Nature. Gharyan, Libya.
*Correspondence: spoterala@gmail.com
Received: 07 November 2023; returned for review: 18 January 2024; accepted: 05 December 2024.
The spur‐thighed tortoise, Testudo graeca, is a widely distributed and deeply diversified species
inhabiting areas of Africa, Asia, and Europe. For decades, populations of T. graeca from North Afri‐
ca have been exploited for the international pet market. In particular, T. graeca of Libyan origin
have been commercially imported into the United States several times since 2021. Using mitochon‐
drial DNA (mtDNA) sequencing, we show that these imported tortoises include T. g. cyrenaica and
a novel lineage originally reported in 2017 from a displaced specimen from a market in Libya. That
novel lineage inhabits northwestern Libya, and we show that wild populations near Gharyan
share the same mtDNA haplotype as those now sold as pets in the USA. Populations of T. graeca in
northwestern Libya, previously identified as T. g. nabeulensis, are reassigned to a new subspecies
Testudo graeca tripolitania described herein. Additional field work is needed to determine the
mtDNA haplotypes of populations in central and southern Tunisia and to locate the subspecies
boundary between T. g. nabeulensis and T. g. tripolitania.
Key words: Africa; holotype; mtDNA; phylogeny; Testudines.

POTERALA ET AL.
2
1758 was actually in Agadir, Morocco, and
had been recorded incorrectly (as being in
Algeria) in historic publications
(Schweiger  Gemel, 2020). To correct this
mistake, T. g. soussensis was re‐designated
as T. g. graeca, and the subspecies in Alge‐
ria and northeastern Morocco (previously
recognized as T. g. graeca) was designated
by the Turtle Taxonomy Working Group
as T. g. whitei (Rhodin et al., 2021). The arri‐
val of T. g. whitei and T. g. nabeulensis into
Europe is comparatively recent, having
occurred in the late Pleistocene (Fritz et al.,
2009; Graciá et al., 2013) and in prehistoric
times (Vamberger et al., 2011; Graciá et al.,
2017a,b).
The range of T. graeca in Asia is largely
contiguous, with a presumption of some
gene flow at subspecies boundaries. The
range of T. graeca in North Africa shows
evidence of both parapatric and allopatric
distribution, with biogeographic barriers
at some subspecies boundaries but also
significant differences in habitat preference
between subspecies (AnadÓn et al., 2015;
Graciá et al., 2017a). Differentiation of the
North African subspecies may have been
driven by oscillation between wet and arid
climates during the mid‐Pleistocene, with
the species’ range being repeatedly frag‐
mented during arid periods (Lambert,
1983; Fritz et al., 2009).
Prior to DNA studies, the identity of
various T. graeca populations was widely
debated in literature, with many new sub‐
species, species, and genera being defined
on the basis of morphological analysis be‐
tween 1986 and 2004 (Chkhikvadze 
Tuniev, 1986; Highfield, 1990; Chkhik‐
vadze  Bakradse, 1991, 2002; Perälä,
1996, 2001, 2002; Pieh, 2000; Weissinger,
2000; Pieh  Perälä 2004; Bonin et al. 2006,
Chkhikvadze et al., 2011). Currently ten
extant subspecies are recognized, and the
monophyletic status of T. graeca has been
firmly established based on mitochondrial
DNA (mtDNA) phylogeny using the cyt-b
gene (Fritz et al., 2007, 2009). Additional
diversity has been proposed, including up
to four distinct sub‐clades of T. g. buxtoni
(Ranjbar et al., 2022).
The present study focuses on a new
North African lineage first reported by
Graciá et al. (2017a). Our initial objective
was to use mtDNA sequencing to identify
a cohort of 28 Libyan Testudo graeca speci‐
mens imported into the United States via
the pet trade. This was based on observa‐
tion that the imported tortoises varied
greatly in appearance and size and the
finding that they could not be accurately
identified morphologically. After finding
that some tortoises represented this un‐
described lineage, a further study was con‐
ducted to geographically locate the line‐
age, provide a detailed physical descrip‐
tion, and assign taxonomic nomenclature.
Our primary goal in this endeavor is to
facilitate conservation efforts for Testudo
graeca in Libya, particularly given the
pressing concern surrounding the sudden
appearance of this lineage in the USA pet
trade. We highlight the vulnerability of
these tortoises to exploitation, alongside
other North African lineages, and empha‐
size that there is a critical need for en‐
hanced protection of these tortoises within
their native range.
Materials and Methods
The captive Libyan spur‐thighed tor‐
toises in this work were studied with the

TESTUDO GRAECA TRIPOLITANIA, A NEW TAXON FROM LIBYA
3
permission of private keepers. All animals
were originally part of commercial impor‐
tations into the United States from Egypt,
which were accompanied by a US Fish &
Wildlife clearance (USFWS Form 3‐177),
indicating that CITES import and export
permits were reviewed. DNA samples
were collected from a total of 28 tortoises
housed within four private facilities in the
United States. Both choanal swabs and
shed epidermis were submied to a com‐
mercial lab (Gendika B.V., Van Ber‐
esteijnstraat 22B, 9641 AB Veendam, Neth‐
erlands) for determination of the cyt‐b
mtDNA haplotypes using sequencing
methods reported in Fritz et al. (2009). Mi‐
tochondrial DNA containing the complete
cyt-b gene and ~20 base pairs (b.p.) of the
adjacent tRNA‐THR gene was amplified
via PCR. The primers used were CytB 5’‐
AAC CAT CGT TGT WAT CAA CTA C‐
3’ (Spinks et al., 2004) and Mt‐E‐Rev2 5’‐
GCR AAT ARR AAG TAT CAT TCT GG‐
3’ (Praschag et al., 2007). Samples were
sequenced using an ABI 3130 (Applied
Biosystems, Foster City, California). Chro‐
matograms were analyzed in UGENE,
checked manually, and were aligned to
published sequences from GenBank
(Clark et al., 2016). Both forward and re‐
verse sequences were analyzed for most
tortoises. We have followed the conven‐
tion of Fritz et al. (2009) in the labeling of
T. graeca cyt-b haplotypes from North Afri‐
ca. Additionally, a total of 44 morphologi‐
cal measurements on the captive tortoises
were collected following the methods of
Pieh  Perälä (2004) to facilitate future
morphological studies. These parameters
are defined in Table 1. Despite veterinary
care, some captive tortoises died during
this study. Three tortoises were preserved
Table 1: Description of morphological measurements collected for Testudo graeca in this study.
ParameterDescriptionParameterDescription
CLMax. carapace lengthV5‐w5th vertebral width
CUCarapace length along curvatureSUP‐lMidline supracaudal length
HEMax. heightSUP‐dMax. supracaudal width (dorsal)
PLMax. plastron lengthSUP‐vMax. supracaudal width (ventral)
PL‐mMidline plastron lengthGU‐lMax. gular length
MIMedian width (marginals 5‐6)GU‐mMidline gular length
MAMax. width (marginals 7‐9)GU‐wMax. combined gular width
NU‐lNuchal lengthGU‐hGular height
NU‐wNuchal widthHUM‐wMax. combined humeral width
C11st costal lengthPEC‐wMax. combined pectoral width
C22nd costal lengthABD‐wMax. combined abdominal width
C33rd costal lengthFEM‐wMax. combined femoral width
C44th costal lengthAN‐wMax. combined anal width
V1‐l1st vertebral lengthHUM‐mMidline humeral seam length (left)
V2‐l2nd vertebral lengthPEC‐mMidline pectoral seam length (left)
V3‐l3rd vertebral lengthABD‐mMidline abdominal seam length (left)
V4‐l4th vertebral lengthFEM‐mMidline femoral seam length (left)

POTERALA ET AL.
4
as wet specimens by fixation in 95% etha‐
nol and storage in 75% ethanol, this meth‐
od being chosen to best preserve DNA for
future studies. These specimens are
housed in the collection of the Field Muse‐
um Chicago under accession numbers
FMNH 289175, 289176, and 289177.
Subsequent to mtDNA analysis of the
captive tortoises, a field study in Libya
was arranged to locate wild T. graeca
matching the novel lineage from Graciá et
al. (2017a). Published range maps indicate
two different ranges for T. graeca in Libya:
in the northwest (Jabal Nafusa mountains
and the adjacent coast of Tripolitania) and
in Cyrenaica (Fritz et al., 2007, 2009; Es‐
Specimen ID
GenBank
Accession
Number
Specimen
Accession
Number
SexCL
(mm)
HE
(mm)
MA
(mm)
WT
(g)Location / Origin
US‐I1‐001*PP942654FMNH 289175F145.786.2111.3649.9
Origin unknown
(Imported to the USA
in 2021)
US‐I1‐002PP942655 F141.378.2105.1574.7
US‐I1‐003PP942656 F136.075.1100.9538.1
US‐I1‐004PP942657 F140.184.1104.7658.1
US‐I1‐005PP942658 F122.475.891.5408.8
US‐I1‐006*PP942659 F122.068.996.2367.9
US‐I1‐007PP942660 M116.967.387.9303.4
US‐I1‐008*PP942661FMNH 289176M110.760.985.4321.1
US‐I1‐009*PP942662FMNH 289177M123.062.092.2353.9
US‐I1‐010PP942663 F152.387.6115.1666.4
US‐I1‐011*PP942664 F145.780.0111.3476.4
US‐I1‐012*PP942665 F142.177.1108.1497.8
US‐I2‐001PP942666 M142.674.2110.3595.3
Origin unknown
(Imported to the USA
in 2022)
US‐I2‐002*PP942667 M112.665.584.1250.0
US‐I2‐003PP942668 M119.573.693.3367.4
US‐I2‐004PP942669 M140.975.098.3480.4
LY‐GRN‐001PP942670 M1206691386
Near Gharyan, Libya
(GPS coordinates
withheld)
LY‐GRN‐002PP942671 F13776100576
LY‐GRN‐003PP942672 M1116280286
LY‐GRN‐004PP942673 M1096280274
LY‐GRN‐005PP942674 F13578106605
LY‐GRN‐006PP942675 F15279109614
LY‐GRN‐007PP942676 F13376101487
 
 
*Deceased
Table 2: Summary of the Testudo graeca belonging to the E lineage included in this study. For pa‐
rameter abbreviation explanation, see Table 1.

TESTUDO GRAECA TRIPOLITANIA, A NEW TAXON FROM LIBYA
5
coriza et al., 2022). The authors made
efforts to inquire within wildlife conserva‐
tion groups in Libya as to whether T. grae-
ca might exist between these two regions
(e.g. along the Gulf of Sirte). Field work
was ultimately conducted in northwestern
Libya, since tortoises in Cyrenaica have
been relatively well sampled and docu‐
mented to be T. g. cyrenaica. Over a period
of one week, a group of six people
(operating as two three‐person teams) was
dispatched to search five sites in Novem‐
ber of 2022. Four sites were within the
Jabal Nafusa Mountains, all within an 8
km radius of Gharyan, Libya. The fifth site
was on an alluvial plain approximately 12
km northwest of Gharyan. These initial
searches did not follow a rigorous method‐
ology, as citizen scientists were recruited
for the effort, and the primary goal was
simply to locate tortoises in their natural
habitat and obtain DNA samples. Due to
time constraints, only basic morphological
data (CL, HE, MA, and WT, see Table 1 for
details), photographs, choanal swabs, and
small samples of loose epidermis were col‐
lected on wild tortoises before they were
released. Mitochondrial DNA analysis of
these tortoises was performed as described
above.
Results
The 2021‐2022 USA imports of Libyan
spur‐thighed tortoises were found to in‐
clude distinct small and large mor‐
photypes. The larger tortoises were gener‐
ally consistent with T. g. cyrenaica as de‐
scribed by Pieh  Perälä (2002), both in
size (CL of ~15‐17 cm) and visual appear‐
ance (in particular, moled carapace
paerning). Some of the imported tortoises
were smaller (~11‐14 cm) and exhibited
more uniform carapace paerning than the
rest, typically without moling. After sub‐
miing samples from both morphotypes to
a commercial lab, we were able to generate
partial (600‐800 b.p.) cyt-b sequences for
each tortoise. The relatively short sequence
lengths reflect poor amplification, which
we primarily aribute to suboptimal sam‐
ple quality (and probably, swab tech‐
nique). Despite multiple aempts at re‐
running samples and analyzing both for‐
ward and reverse sequences, we were una‐
ble to obtain significantly improved results
and made the decision to proceed with the
obtained data. All of the partial sequences
obtained matched exactly either the C1, C2
Figure 1: Wild Testudo graeca and associated
habitat near Tarhunah, Jabal Nafusa moun‐
tains, Libya (Willi Schneider, unpublished
data).

POTERALA ET AL.
6
(T. g. cyrenaica), or E1 (undescribed line‐
age) haplotypes. In particular, out of the 28
tortoises tested among the USA imports,
we identified 16 animals belonging to the E
lineage (Table 2).
Regarding the search for wild tortoises
within Libya, feedback obtained from
wildlife conservation groups was con‐
sistent with the existence of a gap in the
range of T. graeca at the Gulf of Sirte along
the central coast of Libya, as we could find
no reliable records of the species in this
region. As a result, field work was con‐
ducted in northwestern Libya, in regions
where Schneider  Schneider (2008) had
previously reported T. graeca to occur, and
where no DNA sampling had been report‐
ed in the past. Tortoises found by Schnei‐
der  Schneider (2008) and the associated
habitat are shown in Fig. 1 (Willi Schnei‐
der, unpublished data). Using these rec‐
ords and habitat images to inform the
search effort, five sites were searched in
the vicinity of Gharyan, Libya. A total of
seven T. graeca were found at two sites,
both within the Jabal Nafusa mountains at
~700 m elevation. At both sites, tortoise
habitat consisted of flat terrain or gentle
hills with partial vegetation cover. In gen‐
eral, only grasses and herbaceous plants
were present at these sites, including sev‐
eral species of Asteraceae. Interestingly,
the sites where tortoises were found were
the closest in proximity to human develop‐
ment and residential areas surrounding
Gharyan. Of the remaining sites where T.
graeca was not found, two were also within
the Jabal Nafusa Mountains, but at lower
elevations in areas with steeper and more
rocky hillsides. The last site was located on
an alluvial fan northwest of the Jabal
Nafusa Mountains. This site was topo‐
graphically flat but was heavily disturbed
by agricultural activity. Data for the wild
tortoises found are provided in Table 2.
Testudo graeca individuals found near
Gharyan fall within the range of sizes ob‐
served for captive E1 lineage tortoises
from the 2021‐2022 USA imports. As for
the captive specimens, we were able to
recover partial mtDNA cyt-b sequences
(~600‐800 b.p.) for the wild tortoises. For
six of the seven wild tortoises, partial cyt-b
sequences exactly matched the E1 haplo‐
type. The other tortoise (LY‐GRN‐003)
differed from the E1 haplotype by a single
mutation at position 693 of the cyt-b gene,
indicating the presence of a second haplo‐
type within clade E.
Discussion
We agree with Graciá et al. (2017a) that
clade E represents a unique undescribed
lineage, and the present study helps estab‐
lish greater confidence in this finding. In
general, mtDNA phylogeny has given
clear, consistent, and defensible relation‐
ships between Testudo species and clades/
subspecies. In further support to this ap‐
proach, Mikulicek (2013) found nuclear
amplified fragment length polymorphism
clusters to be largely congruent with
mtDNA clades, and Graciá et al. (2017b)
showed similar consistency between
mtDNA results and nuclear
(microsatellite) data for T. g. graeca and T.
g. marrokensis. Phylogenetically, mtDNA
differentiation of the E1 haplotype from
other North African T. graeca subspecies
was shown to be similar to differentiation
between other described subspecies, sup‐
porting a subspecies‐level description

TESTUDO GRAECA TRIPOLITANIA, A NEW TAXON FROM LIBYA
7
(Graciá et al., 2017a).
We have concluded that a detailed
morphological study of this lineage will
require additional field work, as many pri‐
or studies have failed to capture the extent
of morphological variability within each
subspecies. For example, Escoriza et al.
(2022) reported that T. g. nabeulensis exhib‐
its extreme variation in size, with popula‐
tions of larger specimens (CL up to 24.8
cm) not being noted in many prior studies.
The average CL of tortoises in this study
(males 12.1 cm, females 13.9 cm) was com‐
parable to that of T. g. nabeulensis from
northern Tunisia (males 12.1 cm, females
13.0 cm) studied by Pieh  Perälä (2002).
Populations near Sfax, Tunisia (Escoriza et
al., 2022) were even smaller (males 11.1 cm,
females 12.1 cm), but without mtDNA test‐
ing we cannot be certain that these tortois‐
es are T. g. nabeulensis. The E lineage tor‐
toises in this study were very high‐domed
in profile, with an average (± standard de‐
viation) normalized shell height (HE/CL)
of 0.56 ± 0.028.
We hypothesize that E lineage T. graeca
are geographically isolated from both T. g.
cyrenaica and T. g. nabeulensis. The mixing
with T. g. cyrenaica in import groups al‐
most certainly occurred in captivity, as the
small range of T. g. cyrenaica has been well
established and thoroughly sampled with‐
out the discovery of clade E haplotypes in
wild tortoises. Likewise, no T. g. nabeulen-
sis (clade A) haplotypes were found in ei‐
ther the wild tortoises from the Jabal
Nafusa mountains or the 2021‐2022 USA
imports. Based on these factors, we infer
that a significant wild population of E line‐
age spur‐thighed tortoises exists and rep‐
resents a new North African subspecies of
T. graeca. We found no evidence of a prior
description of the lineage. The name Testu-
do flavominimaralis (Highfield  Martin,
1989), not recognized as valid, was as‐
signed to a displaced tortoise of North Af‐
rican origin. The identity of this specimen
is uncertain, as it lacks a type locality and
has a considerably lower shell profile (HE/
CL = 0.487) than the clade E tortoises in
this work. Clade E is described herein:
Testudo graeca tripolitania n. subsp.
Tripolitanian Tortoise
Holotype – An adult male (US‐I1‐008)
with the E1 cyt-b haplotype, imported into
the United States in 2021, and believed to
originate from northwestern Libya. The
holotype specimen, shown in Fig. 2, is typ‐
ical in size, appearance, and morphology
for the subspecies, except for having scute
anomalies. Specifically, the nuchal is miss‐
Figure 2: Holotype of Testudo graeca tripolitania
n. subsp.

POTERALA ET AL.
8
ing (typically present but varies in size)
and the supracaudal scute is divided
(more commonly undivided). Scute anom‐
alies are common in T. graeca and should
not be regarded as diagnostic (Mira‐Jover
et al., 2024). The holotype died of unknown
causes in 2022 and was fixed in 95% etha‐
nol and preserved in 75% ethanol. The hol‐
otype is stored at the Field Museum of
Natural History as accession number
FMNH 289176.
Paratypes – A total of 15 additional tor‐
toises (seven males and eight females)
were sampled from USA commercial im‐
portations in 2021 and 2022 and were veri‐
fied to have the same cyt-b haplotype (E1)
as the holotype specimen. Male US‐I1‐009
and female US‐I1‐001 died during the
course of this study, and are preserved as
wet specimens at the Field Museum
(FMNH 289177 and 289175, respectively).
Four additional tortoises (females US‐I1‐
006, US‐I1‐011, US‐I1‐012, and male US‐I2‐
002) died in captivity, and the remaining
tortoises are still living in private facilities
in the United States. A subset of these tor‐
toises are shown in Fig. 3 to exemplify typ‐
ical variability in coloration and paern.
For comparison, wild tortoises (three
males and four females) found near Ghar‐
Figure 3: Testudo graeca tripolitania n. subsp. Females (left pictures) and males (right pictures) im‐
ported into the USA in 2021‐2022.

TESTUDO GRAECA TRIPOLITANIA, A NEW TAXON FROM LIBYA
9
yan, Libya, are shown in Fig. 4. These tor‐
toises are also morphologically consistent
with the holotype, not considering scute
anomalies. Six of these wild tortoises share
the E1 cyt-b haplotype, and the seventh is
identified as belonging to clade E but hav‐
ing a previously unreported haplotype.
Description and Comparison – The car‐
apace is rounded in appearance and very
high‐domed in profile, with a yellow‐tan
or yellow‐orange base color and bold black
paerning. The typical paern consists of a
central black spot on each vertebral or cos‐
tal scute, combined with a black rim along
the anterior and lateral edges of each ver‐
tebral scute and along the anterior (and
sometimes ventral) edges of each costal
scute. Typically, black pigment also fills a
triangular region between the areola and
the anterior edge of most marginal scutes.
The plastron is yellow‐tan with sharply
contrasting and irregular black paerning.
The head and limbs are predominantly
yellow with intermixed black pigment,
particularly on the top of the head and
forelimb scales. Complete morphological
parameters are given in Table S1. All tor‐
toises appear mature with no evidence of
recent growth. We also examined the rela‐
tionship between CL and HE/CL for North
African T. graeca (Fig. 5). Comparative data
for other North African subspecies were
derived from Highfield (1990), Pieh
(2000), Pieh  Perälä (2002, 2004) and Tiar
‐Saadi et al. (2022) and are presented as
average values for males and females in
Figure 4: Testudo graeca tripolitania n. subsp.
from sites near Gharyan, Libya. Pictures on the
left show individuals from site #1; pictures
above show individuals from site #2.

POTERALA ET AL.
10
each studied population. Averages are
used because most authors have elected to
withhold morphology data on individual
tortoises. For T. g. tripolitania n. subsp., we
show separate averages for the 16 captive
tortoises and for the seven wild tortoises
from Gharyan, Libya.
Genetic differentiation between T. g.
tripolitania n. subsp. and other recognized
T. graeca subspecies was previously dis‐
cussed by Graciá et al. (2017a). Support for
differentiation of T. g. tripolitania n. subsp.
was found to be similar to that for other
North African subspecies, and this lineage
was found to have diverged from T. g.
graeca during the Pleistocene. The domi‐
nance of a single cyt-b haplotype within
the population may indicate low genetic
diversity relative to other T. graeca subspe‐
cies, though we do find evidence that at
least one additional haplotype exists, and
more diversity may yet be discovered by
further range surveys.
Remarks – Morphological diagnosis of
subspecies in T. graeca is fraught with diffi‐
culty, as evidenced by an extensive list of
now‐invalidated genera, species, and sub‐
species. Despite this, there are distinct
morphological characteristics of T. g. tripo-
litania n. subsp., which are useful for diag‐
nosis. These tortoises are very small in
size, similar to the smallest documented T.
g. nabeulensis and T. g. whitei. In addition,
the subspecies exhibits a very high shell
profile and has minimal flaring of the mar‐
ginal scutes. The carapace is rounded in
profile and is highest at the 3rd vertebral
scute, with a dome‐shaped appearance.
The shell is wider toward the posterior
side, typically being widest at the 9th mar‐
Figure 5: Plot of normalized shell
height (HE/CL) versus straight cara‐
pace length (CL) for North African
Testudo graeca subspecies. Each sym‐
bol represents the average of a study
population (captive and wild groups
of T. g. tripolitania n. subsp. are indi‐
cated)

TESTUDO GRAECA TRIPOLITANIA, A NEW TAXON FROM LIBYA
11
ginal scute. The carapace has a yellow, yel‐
low‐tan, or yellow‐orange base color with
highly contrasting black paerning (Figs.
3, 4). Most individuals show a well‐
defined paern like the holotype, but in
some cases the black pigmentation may be
irregular or show moling as in T. g. cyre-
naica. The plastron is typically yellow with
irregular and asymmetric black pigmenta‐
tion, but may be predominantly black in
some individuals. The head may range
from nearly entirely yellow to mostly
black, while the forelimbs are predomi‐
nantly yellow with occasional black leg
scales. While some T. g. cyrenaica exhibit
similar coloration, adults can typically be
distinguished by differences in size and
HE/CL ratios. These tortoises are difficult
to distinguish from T. g. nabeulensis and
T.g. whitei, as tortoises from Algeria or the
northern Tunisian coast are very similar in
both size and appearance (Highfield, 1990;
Rhodin et al., 2021; Tiar‐Saadi et al., 2022).
Testudo g. terrestris also exhibits well‐
defined yellow and black carapace paern‐
ing and a similar overall appearance, but
typically lacks the characteristic bold pig‐
mentation on the head and forelimbs
(Rhodin et al., 2021).
Etymology – The subspecific name trip-
olitania references the historical region of
Tripolitania in northwestern Libya, where
the tortoises are native. Tripolitania is de‐
rived from Greek, meaning “three cities”.
Type Locality – Northwestern Libya. In
the absence of exact locality data for the
holotype, we refer to ICZN Art 76.1.1, “If
capture or collection occurred after
transport by artificial means, the type lo‐
cality is the place from which the name‐
Figure 6: Range for Testudo graeca in eastern Algeria, Tunisia, and Libya. Filled symbols indicate
localities for T. g. whitei, nabeulensis, cyrenaica, and tripolitania n. subsp. with mtDNA confirmation,
including data from Fritz et al. (2009), and AnadÓn et al. (2015). Black dots indicate records with‐
out mtDNA confirmation from AnadÓn et al. (2015), Schneider  Schneider (2008), and Rhodin et
al. (2021). The cross‐hatched region indicates areas where T. graeca occurs but the subspecies iden‐
tity is uncertain. The single‐hatched region indicates areas where the presence of T. graeca is un‐
certain.

POTERALA ET AL.
12
bearing type, or its wild progenitor, began
its unnatural journey.” We have collected
evidence to show that populations in
northwestern Libya are the wild progeni‐
tor of the type specimen, with populations
near Gharyan being consistent genetically
and morphologically with the type. The
presence of spur‐thighed tortoises in
northwestern Libya was reported by High‐
field (1990) and was further documented
by Schneider  Schneider (2008). These
tortoises were historically assigned to T. g.
nabeulensis based on appearance, and this
interpretation was followed by other au‐
thors investigating mtDNA phylogeny of
the species, including Fritz et al. (2007,
2009), despite being approximately 400
kilometers from the nearest mtDNA veri‐
fied records in northern Tunisia.
Range and Ecology – Additional field
work is necessary to confidently establish
the range of T. g. tripolitania n. subsp. Be‐
cause previous range maps suggest a con‐
tiguous range for T. graeca from Tunisia to
Libya (Fritz et al., 2009; Rhodin et al.,
2021), the location of a subspecies bounda‐
ry is unclear. An updated map for T. graeca
in Libya, Tunisia, and western Algeria is
proposed in Fig. 6. We restrict identifica‐
tion of subspecies to regions where it is
well supported by mtDNA sampling. Con‐
sequently, there is a large region where T.
graeca is present but not adequately identi‐
fied: from Sfax, Tunisia, south to Ta‐
taouine, and west to approximately 100
kilometers beyond the Algerian border.
We also identify regions where T. graeca
has not been recorded but may plausibly
occur, particularly between Tataouine, Tu‐
nisia and the Jabal Nafusa mountains in
Libya. If allopatric speciation is consid‐
ered, there are two plausible locations for a
biogeographic barrier between T. g. na-
beulensis and T. g. tripolitania n. subsp. The
first is near the salt lake Cha el Djerid in
central Tunisia, where low elevation and
lack of vegetation limit tortoise habitat to a
narrow strip along the Gulf of Gabes. The
second is near the Tunisia‐Libya border,
another area of lower elevation where T.
graeca has so far not been reported. We
also consider the strong evidence for niche
partitioning in North African T. graeca re‐
ported by AnadÓn et al. (2015). The known
habitat of T. g. tripolitania n. subsp. in the
Jabal Nafusa range is relatively warm and
dry, with low vegetation cover and annual
precipitation of 150‐250 mm. This habitat
contrasts with that of T. g. nabeulensis in
northern Tunisia, where annual precipita‐
tion is 300‐1200 mm and vegetation cover
is higher. Potential T. graeca habitat in cen‐
tral and southern Tunisia (from the Gulf of
Gabes south to the Libyan border) is more
similar to that of Gharyan, Libya, than to
northern Tunisia, and we find plausible
that the range of T. g. tripolitania n. subsp.
could extend into Tunisia.
Conservation Remarks – Our foremost
concern is the ongoing importation of a
new lineage of T. graeca in the USA pet
trade, along with T. g. cyrenaica, and the
dubious nature of these imports. These T.
graeca shipments have been consistently
approved for legal commercial import into
the USA despite serious reason for con‐
cern, namely, their export from a country
where they are not native (Egypt) but
which is geographically adjacent to their
native range in Libya. Illegal transport of
T. graeca from Libya to Egypt is frequent,
as personally recounted by authors of this

TESTUDO GRAECA TRIPOLITANIA, A NEW TAXON FROM LIBYA
13
work who have assisted Libyan authorities
in caring for and releasing confiscated tor‐
toises. We urge CITES to ensure that inter‐
national trade of North African T. graeca is
legal and biologically sustainable, as popu‐
lations of these tortoises are at risk of de‐
pletion without oversight and enforcement
of existing laws. Moreover, we request that
USFWS considers the concerns raised here‐
in when reviewing any future importation
of T. graeca.
References
AnadÓn, J.D.; Graciá, E.; Botella, F.; Giménez,
A.; Fahd, S.  Fritz, U. (2015). Individualis‐
tic response to past climate changes: niche
differentiation promotes diverging Quater‐
nary range dynamics in the subspecies of
Testudo graeca. Ecography 38: 956‐966.
Bonin, F.; Devaux, B.  Dupré, A. (2006). Turtles
of the World. Johns Hopkins University
Press, Baltimore, USA.
Chkhikvadze, V.M.  Bakradze, M.A. (1991).
[On the taxonomic status of the recent land
turtle from the Araxes valley]. Proceedings of
the Tbilisi State University 305: 59‐63. (in Rus‐
sian)
Chkhikvadze, V.M.  Bakradze, M.A. (2002). [A
new species of land tortoises of Dagestan].
Proceedings of the Zoology Institute of the Geor-
gia Academy of Sciences 21: 276‐279. (in Rus‐
sian)
Chkhikvadze V.M.  Tuniev B.S. (1986). [On the
taxonomic status of the modern land tor‐
toise of Western Transcaucasus]. Bulletin of
the Academy of Sciences of the Georgian SSR
124: 617‐620. (in Russian)
Chkhikvadze V.M.; Mazanaeva L.F.  Shamma‐
kov S.M. (2011). [A short account of a new
species of land tortoise in Dagestan], In Pro-
ceedings of the International Conference on
Biological Diversity and Conservation Problems
of the Fauna of the Caucasus. AUA Acopian
Center for the Environment, Erevan, Arme‐
nia, pp. 336‐340. (in Russian)
Clark, K.; Karsch‐Mizrachi, I.; Lipman, D.J.;
Ostell, J.  Sayers, E.W. (2016). GenBank.
Nucleic Acids Research 44: D67‐D72.
Escoriza, D.; Díaz‐Paniagua, C.; Andreu, A. 
Ben Hassine, J. (2022). Testudo graeca Lin‐
naeus 1758 (Western Subspecies Clade: Tes-
tudo g. graeca, T. g. cyrenaica, T. g. marokken-
sis, T. g. nabeulensis, T. g. whitei) – Mediterra‐
nean Spur‐thighed Tortoise, Moorish Tor‐
toise, Libyan Tortoise, Moroccan Tortoise,
Tunisian Tortoise, Souss Valley Tortoise.
Chelonian Research Monographs 5: 117.1‐
117.18.
Fritz, U.; HundsdÖrfer, A.K.; ŠirokÝ, P.; Auer,
M.; Kami, H.; Lehmann, J.; Mazanaeva,
L.F.; TÜrkozan, O.  Wink, M. (2007). Phe‐
notypic plasticity leads to incongruence
between morphology‐based taxonomy and
genetic differentiation in western Palaearc‐
tic tortoises (Testudo graeca complex; Testu‐
dines, Testudinidae). Amphibia-Reptilia 28:
97‐121.
Fritz, U.; Harris, D.J.; Fahd, S.; Rouag, R.;
Graciá Martínez, E.; Giménez Casalduero,
A.; ŠirokÝ, P.; Kalboussi, M.; Jdeidi, T.B. 
Hundsdorfer, A.K. (2009). Mitochondrial
phylogeography of Testudo graeca in the
Western Mediterranean: old complex diver‐
gence in North Africa and recent arrival in
Europe. Amphibia-Reptilia 30: 63‐80.
Graciá, E.; Giménez, A.; AnadÓn, J.D.; Harris,
D.J.; Fritz, U.  Botella, F. (2013). The un‐
certainty of Late Pleistocene range expan‐
sions in the western Mediterranean: a case
study of the colonization of south‐eastern
Spain by the spur‐thighed tortoise, Testudo
graeca. Journal of Biogeography 40: 323‐334.
Graciá, E.; Vargas‐Ramírez, M.; Delfino, M.;
AnadÓn, J.D.; Giménez, A.; Fahd, S.; Corti,
C.; Jdeidi, T.B.  Fritz, U. (2017a). Expan‐
sion after expansion: dissecting the phylo‐
geography of the widely distributed spur‐
thighed tortoise, Testudo graeca (Testudines:
Testudinidae). Biological Journal of the Linne-
an Society 121: 641‐654.
Graciá, E.; Rodriguez‐Caro, R.C.; Andreu,

POTERALA ET AL.
14
A.C.; Fritz, U.; Gimenez, A.  Botella, F.
(2017b). Human‐mediated secondary con‐
tact of two tortoise lineages results in sex‐
biased introgression. Scientific Reports 7:
4019.
Highfield, A.C. (1990). Tortoises of North Afri‐
ca: Taxonomy, nomenclature, phylogeny,
and evolution with notes on field studies in
Tunisia. Journal of Chelonian Herpetology 1: 1‐
56.
Highfield, A.C.  Martin, J. (1989). Description
of a Miniature Tortoise Testudo flavomin‐
imaralis n. species from north Africa. The Tor‐
toise Trust, London, UK.
Lambert, M.R.K. (1983). Some factors influenc‐
ing the Moroccan distribution of the west‐
ern Mediterranean spur‐thighed tortoise,
Testudo graeca graeca L., and those preclud‐
ing its survival in NW Europe. Zoological
Journal of the Linnean Society 79:149‐178.
Mikulicek, P.; Jandzik, D.; Fritz, U.; Schneider,
C.  ŠirokÝ, P. (2013). AFLP analysis shows
high incongruence between genetic differ‐
entiation and morphology‐based taxonomy
in a widely distributed tortoise. Biological
Journal of the Linnean Society 108: 151‐160.
Mira‐Jover, A.; Rodríguez‐Caro, R.C.; Nogue‐
ra, J.C.; Fritz, U.; Kehlmaier, C.; García de
la Fuente, M.I.; Giménez, A.  Graciá, E.
(2024). Shorter telomeres are associated
with shell anomalies in a long‐lived tortoise.
Journal of Zoology 323: 67‐75.
Perälä, J. (1996). [Turtles of southern Turkey.
Morphological and ecological differences
(Testudo ibera Pallas 1841 & Testudo ibera
anamurensis Weissinger 1987) and the de‐
scription of a new turtle species]. In Herpe-
tokongressi I, Appendix. Helsinki, Finland,
pp. 14‐26. (in Finnish)
Perälä, J. (2001). The genus Testudo
(Testudines: Testudinidae): Phylogenetic
inferences. Chelonii 3: 32‐39.
Perälä, J. (2002). Morphological variation
among Middle Eastern Testudo graeca L.,
1758 (sensu lato) with a focus on taxonomy.
Chelonii 3: 78‐108.
Pieh, A. (2000). Testudo graeca soussensis, eine
neue Unterart der Maurischen Land‐
schildkröte aus dem Sousstal (Südwest‐
Marokko). Salamandra 36: 209‐222.
Pieh, A.  Peräla, J. (2002). Variabilität von
Testudo graeca Linnaeus, 1758 im östlichen
Nordafrika mit Beschreibung eines neuen
Taxons von der Cyrenaika (Nordostlibyen).
Herpetozoa 15: 3‐28.
Pieh, A.  Peräla, J. (2004). Variabilität der
Maurischen Landschildkröten (Testudo grae-
ca Linnaeus, 1758 – Komplex) im zentralen
und nordwestlichen Marokko mit
Beschreibung zweier neuer Taxa. Herpetozoa
17: 19‐47.
Praschag, P.; HundsdÖrfer, A.K.  Fritz, U.
(2007). Phylogeny and taxonomy of endan‐
gered South and Southeast Asian freshwa‐
ter turtles elucidated by mtDNA sequence
variation (Testudines: Geoemydi‐
dae: Batagur, Callagur, Hardella, Kachuga,
Pangshura). Zoologica Scripta 36: 429‐442.
Ranjbar, N.; Malekian, M.; Ashrafzadeh, M.R.
 Hemami, M.R. (2022). Phylogeographic
and phenotypic divergence between two
subspecies of Testudo graeca (T. g. buxtoni
and T. g. zarudnyi) across their contact zone
in Iran. Scientific Reports 12: 13579.
Rhodin, A.; Iverson, J.; Bour, R.; Fritz, U.;
Georges, A.; Shaffer, H.  Dijk, P.P. (2021).
Turtles of the World: Annotated Checklist
and Atlas of Taxonomy, Synonymy, Distri‐
bution, and Conservation Status, 9th ed.
Chelonian Research Foundation and Turtle
Conservacy, IUCN, Gland, Swierland.
Schneider, C  Schneider, W. (2008). The
Egyptian Tortoise, Testudo kleinmanni Lortet,
1883 in Libya. Salamandra 44: 141‐152.
Schweiger, M.  Gemel, R. (2020). Where do
you come from, stranger? A scientific histor‐
ical digression with discussion on nomen‐
clature and taxonomy of Testudo graeca Lin‐
naeus, 1758. Herpetozoa 33: 31‐38.
Spinks, P.Q.; Shaffer, H.B.; Iverson, J.B. 
McCord, W.P. (2004). Phylogenetic hypoth‐
eses for the turtle family Geoemydidae.

TESTUDO GRAECA TRIPOLITANIA, A NEW TAXON FROM LIBYA
15
Molecular Phylogenetics and Evolution 32: 164‐
182.
Tiar‐Saadi, M.; Tiar, G.; Bouslama, Z.  Ši‐
rokÝ, P. (2022). Mechanisms Determining
body size and shape difference in Algerian
spur‐thighed tortoises (Testudo graeca). Ani-
mals 12: 1330.
TÜrkozan, O.; Javanbakht, H.; Mazanaeva, L.;
Meiri, S.; Kornilev, Y.V.; Tzoras, E.; Pop‐
georgiev, G.; Shanas, U.  Escoriza, D.
(2023). Testudo graeca Linnaeus 1758
(Eastern Subspecies Clades: Testudo g. arme-
niaca, Testudo g. buxtoni, Testudo g. ibera,
Testudo g. terrestris, Testudo g. zarudnyi) –
Armenian Tortoise, Zagros Tortoise, Anato‐
lian Tortoise, Levantine Tortoise, Kerman
Tortoise. Chelonian Research Monographs 5:
120.1‐120.33.
Vamberger, M.; Corti, C.; Stuckas, H.  Fritz,
U. (2011). Is the imperilled spur‐thighed
tortoise (Testudo graeca) native in Sardinia?
Implications from population genetics and
for conservation. Amphibia-Reptilia, 32: 9‐25.
Weissinger, H. (2000). Testudo graeca anamuren-
sis ssp. nov. from Asia Minor. Asiatic Herpe-
tological Research 9:145‐148.