TANNUASPIS, TUVASPIS A N D ILEMORASPIS,

ENDEMIC OSTEOSTRACAN GENERA FROM THE SILURIAN AND DEVONIAN

OF T U V A A N D K H A K A S S I A (USSR)

by
OLGA AFANASSIEVA * & PHILIPPE JANVIER **

ABSTRACT RI~SUMfl

Three osteostracan genera, Tannuaspis, Tuvaspis
and Ilemoraspis, from the Silurian and Devonian of
the Autonomous Republic of Tuva and the Autono-
mous Region of Khakassia (USSR), are revised on the
basis of the type material briefly described by Obru-
chev (1956, 1961, 1964). Tannuaspis bears some
resemblances with the Tremataspididae. Tuvaspis
may be closely related to Tannuaspis. Finally, there is
no evidence of paired fins and cornual processes in
Ilemoraspis (neither is there any evidence of their
absence). Tannuaspis and Ilemoraspis share some
characters which are not found in any other osteostra-
can: widely separated orbits, very small and oval
lateral fields, situated at the same level as the median
dorsal field. These resemblances may suggest that
these two genera (to which Tuvaspis may be added)
form a monophyletic group of osteostracans, which
may be endemic to this part of Central Asia.
Trois genres d'Ost6ostrac6s, Tannuaspis, T.uvasp&
et Ilemoraspis, du Siluro-D6vonien de la R6publique
Autonome de Tuva et de la R6gion Autonome de
Khakass (URSS), sont r6vis6s h partir du mat6riel
type, bri~vement d6crit par Obruchev (1956, 1961,
1964). Tannuaspis se rapprocherait, par certains
caract6res, des Tremataspididae. Tuvaspis est proba-
blement tr6s proche de Tannuaspis. Enfin, il n'y a pas
de certitude quant ~ la pr6sence de nageoires paireset
de processus cornuaux chez Ilemoraspis. Par ailleurs,
Tannuaspis et llemoraspis partagent certains caract6-
res que l'on ne trouve chez aucun autre Ost6ostrac6 :
orbites tr~s espac6es, champs lat6raux tr6s petits, ova-
les, et situ6s au niveau du champ m6dian dorsal. Ces
ressemblances singulibres sugg6reraient que ces deux
genres (auxquels on peut adjoindre Tuvaspis) consti-
tuent un groupe monophyl6tique d'Ost6ostrac6s,
end6miques ~t cette r6gion de l'Asie centrale.

KEY-WORDS : OSTEOSTRACI (VERTEBRATA), SILURIAN-DEVONIAN, CENTRAL ASIA (USSR), REVISION, SYSTEMATICS,

BIOGEOGRAPHY.

MOTS-CLI~S : OSTEOSTRACI (VERTEBRATA), SILURO-DI~VONIEN, ASIE CENTRALE (URSS), RI~VISION, SYSTI~MATIQUE~

BIOGI~OGRAPHIE.

* Paleontological Institute of the Academy of Sciences of the USSR, Profsoyouznaya 113, Moscow, USSR.
** LA 12 du CNRS, Institut de Pal6ontologie, 8, rue Buffon, 75005 Paris, France.

Geobios, n ° 18, fasc. 4 p. 493-506, 7 figs. Lyon, aofit 1985

 - - 494 - -

CONTENTS

Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . p. 494 3.1.1. Genus TannuaspisOBRUCHEV,1956. p. 497

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . p. 494 3.1.2. Genus TuvaspisOBRUCHEV,1956 . . . p. 501

2. Geological setting . . . . . . . . . . . . . . . . . . . . . . p. 495

3.2. Genus llemoraspis OBRUCHEV,1961 . . . p. 501
4. Conclusions . . . . . . . . . . . . . . . . . . ~. . . . . . . . p. 505
3. Description and discussion . . . . . . . . . . . . . . p. 495
3.1. Tannuaspididae OBRUCHEV, 1964 . . . . . p. 495 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . p. 505

Abbreviations - Abr6viations

PIAN : Paleontological Institute of the Academy oe.r., occipital region, r6gion occipitale.
of Sciences (Akademia Nauk) of the USSR, Moscow. orb., orbit, orbite.
br.f., branchial fossa, fosse branchiate. or.n., oral notch, 6chancrure orale.
br.n., branchial notch, 6chancrure branchiale. pi.f., pineal foramen, foramen pin6al.
er., crack, cassure. p.lim.end., posterior limit of endoskeleton, limite
exbr., extrabranchial division of branchial fossa, post6rieure de l'endosquelette.
division extrabranchiale de la fosse branchiale.
sc., scales, 6cailles.
ibr., interbranchial ridge, cr~te interbranchiale.
s.e.l., canal leading to the lateral field, canal con-
i.orb., i n f r a o r b i t a l sensory-line, ligne infra- duisant au champ lat6ral.
orbitaire.
tes., tesserae, tess6res.
l.f., lateral field, champ lat6ral.
transv., transverse sensory line, ligne transverse.
m.d.f., median dorsal field, champ m6dian dorsal.
tub., tubercles, tubercules.
m.d.er., median dorsal crest, cr~te m6diane dorsale.
v.i.l., ventrolateral endoskeletal lamina, lame
nhyp., naso-hypophysial opening, orifice naso-
hypophysaire. endosquelettique ventro-lat6rale.
o b r . e . , o r a l o b r a n c h i a l cavity, cavit6 oralo- x., slit-like opening of unknown function, ouver-
branchiale. t u r e e n forme de fente et de fonction inconnue.

1. I N T R O D U C T I O N

The osteostracans discussed in the present paper
were discovered in Tuva and Khakassia by geologists
in 1952 and 1954, and they still are the only known
evidence of this group of jawless vertebrates in Cen-
tral Asia. Both regions are situated in the South of
Siberia. The Autonomous Republic of Tuva stretches
between the southern part of the Krasnoyarsk region
and the northern border of Mongolia. The Autono-
mous Region of Khakass lies southwest of the Kras-
noyarsk region and along the northern border of the
A.R. of Tuva.
This material has been briefly described by D.V.
Obruchev (1956, 1961, 1964), who erected the three
generic names Tannuaspis, Tuvaspisand Ilemoraspis.
New investigations on this material have yielded the
results presented herein, and a consideration of the
phylogenetic relationships of these genera is propo-
sed.
 -- 495 --
2. GEOLOGICAL SETTING
The territory of the A.R. of Tuva has yielded seve-
ral osteostracan-bearing localities briefly mentioned
by Obruchec (1956, 1961). Although these localities
are scattered over the whole territory, they are litholo-
gically quite similar.
The osteostracan remains are found within a thick
section beginning with limestones of Wenlockian age
and ending just below an Eifelian level. The age of
these osteostracans may thus be either Uppermost
Silurian or Lower Devonian.
The remains referred to the genus Tannuaspis are
met with in the Chondergeyskaya, Samagaltayskaya
and Kendeyskaya Group of the Tuvin depression in
southwestern Tuva (Obruchev, 1961, 1964 ; Melests-
henko & alii, 1973).
The Chondergeyskaya Group is referred to the
Upper Silurian and consists of red sandstone and
aleurolites, occasionally conglomerates and gravels.
Its average thickness is 2000-3000 m (Vladimirskaya,
1975). Besides the osteostracan remains, it contains
rare remains of brachiopods and ostracods. This
series passes gradually to the overling Samagaltays-
kaya Group, which is uncomformably overlain by the
Kendeyskaya Group. These two latter are of Lower
Devonian age.
The Samagaltaskaya Group consists of red sands-
tone and aleurolites, with interbedded tuffaceous
rocks (Predtechenskiy, 1975), and its average thick-
ness is 1000 m. This group also contains remains of
osteostracans, associated with ostracods and gastro-
pods.
The Kendeyskaya Group consists of various volca-
nic rocks, interbedded with red, sometimes many-
coloured sandstones and aleurolites. It passes gra-
dually to the overlying Saglinskaya Group of Lower
Eifelian age. Its thickness ranges from 1000 m to 3000
m and it has yielded abundant remains of osteostra-
cans, acanthodians, brachiopods and gastropods. The
remains referred by Obruchev (1956) to Tannuaspis,
including the holotype of T. levenkoi OBRUCHEVhave
3. DESCRIPTION AND DISCUSSION
3.1. Tannuaspididae OBRUCHEV,1964.
This family is represented by Tannuaspis levenkoi
OBRUCHEV (1956) and probably also Tuvaspis marga-
ritae OBRUCHEV (1956), a pporly known species from
been discovered in this group, in a locality situated on
the eastern slope of the Kutuk dry valley (on the sou-
thern slope of the western Tannu-Ola mountains).
There, only osteostracan remains occur in a very hard
reddish sandstone.
The remains referred to Tuvaspis (Obruchev, 1964)
are from the Samagaltayskaya Group and have been
found mainly southwest of Kizil, near Lake Khadin.
They are mainly represented by fragmentary material.
Obruchev (1956) emphasized that, since the Tan-
nuaspis and Tuvaspis remains are found throughout
the whole thickness of these groups, it is difficult to
assign them to any precise age. Moreover, the inverte-
brate fragments found in association with them are
practically undeterminable. Obruchev made a tenta-
tive estimation of the age of Tannuaspis on the basis
of morphological characters, such as the size of the
lateral fields, which he considered as being quite
large. However, our investigations show that some of
the date used by this author for this purpose were not
accurate and, thus his conclusions may be modified
accordingly.
The remains referred to Ilemoraspis are from the
Ilemorovskaya Group and were collected in a locality
situated in the Tchazy-Koiza area, in Khakassia
(Obruchev, 1961, 1964). This group is usually referred
to the Givetian (Obruchev, 1956; Melestshenko &
alii, 1973 ; Rzonsnitskaya & Maymina, 1975), howe-
ver, Obruchev (1964) later referred Ilemoraspis to the
Lower Devonian without giving any particular rea-
son.
According to Rzonsnitskaya & Maymina (1975), the
Ilemorovskaya Group consists of grey sandstones,
tuffstones, aleurolites and cherts, and its thickness
ranges from 84 m to 550 m. Besides Ilemoraspis, this
group has yielded remains of brachiopods, cirripeds
and phyllopods, including the typically Middle Devo-
nian species Pseuclestheria pogrebovi LUTK., which is
characteristic of the Givetian Narova beds of the
Main Devonian Field (Melestshenko & alii, 1973).
the Silurian or Devonian of Tuva. These two species
have been briefly described by Obruchev (1956, 1961,
1964), who did not relate them to any particular
group of the West European osteostracans. However,
this author compared Tannuaspis to the Tremataspi-
 -- 496 --

vl.I._ ~
im.en

B2

Fig. 1 - - A, Tannuaspis levenkoi OaRUCnEV, Silurian-Devonian, Tuva, USSR, Holotype, P I A N n ° 1013/1. Rubber cast of the specimen
(A1) and interpretation (A2) ; exoskeleton of the ventral surface of the shield in grey. B, Tannuaspis ef. levenkoi OaRUCrIEV, same
age and locality as for A, P I A N n ° 1013/2. Rubber cast of the specimen (B1) and interpretation (B2). Scale : 1 cm.
A, moulage en 61astom6re du sp6cimen (A1) et son interpr6tation (A2) ; en gris6 : exosquelette de la face ventrale du bouclier. B,
moulage en 61estom6re du sp6cimen (B1) et son interpr6tation (B2). l~chelle : 1 cm.
 - - 497 --
didae, a group of thyestidian osteostracans lacking
paired fins (Janvier , 1981b, 1984, 1985). These two
species deserve a more detailed description, with
regard to the general problem of osteostracan interre-
lationships.
3.1.1. Genus Tannuaspis OBRUCHEV,1956
Tannuaspis levenkoi OBRUCHEV, 1956
(Figs. 1, 2, 3, 4A, B)
This species is represented by the holotype (PIAN,
1013/1), preserved mainly as a natural cast of the
shield with fragments of remaining exoskeleton, in a
very hard quartzitic sandstone, and a second speci-
men from the same locality (PIAN 1013/2). The latter
shows the dorsal exoskeleton, partly exposed in ven-
tral aspect, and part of the dorsal impression of the
exoskeleton (Figs. 1B, 3). It is noteworthy that this
second specimen is clearly larger and displays somew-
hat different proportions than the holotype. More
material would be necessary, however, to decide whe-
ther it belongs to the range of variation of T. levenkoi
or represents a different species. The two specimens
will be described here separately.
"k" THE HOLOTYPE (Figs. 1A, 2).
This specimen is preserved in dorsal view, the dorsal
exoskeleton being removed. The natural internal cast
of the oralobranchial and abdominal cavities are visi-
ble and, laterally, a part of the impression of the
external surface of the ventral exoskeleton is exposed
(stippled in Fig. 1A). A trace of the orbital cavities is
visible (orb., Fig. 1A1), but the presumed naso-
hypophysial region is badly preserved, and no naso-
hypophysial opening is visible. Neither is any pineal
foramen visible. The cast of the oralobranchial cavity
shows traces of, at least, three or four interbranchial
ridges (ihr., Figs. 1A1, 2B), which separate relatively
large branchial fossae (br.f., Fig. 2B). The latter pos-
sess a lateral, extrabranchial division (exbr., Fig. 2B),
ending laterally with a branchial notch surrounded by
a slight exoskeletal thickening (br.n, Fig. 2B).
The natural cast of the dorsal part of the abdominal
cavity shows the possible posterior limit of the endos-
keleton (p.lim.end., Figs. IA2, 2C). The internal sur-
face of the abdominal cavity which was lined with
endoskeleton is roughened, whereas that covered only
with exoskeleton is smooth. This reveals that the
endoskeletal occipital region (oc.r., Fig. 2C) was very
broad, and that a ventrolateral endoskeletal lamina
(v.l.l., F i g . 1A2) extended posteriorly as far as the
posterior limit of the shield. Contrary to previous
reconstructions (Obruchev, 1956, fig. 2 ; 1961, fig.
92 ; 1964, fig. 30), there is no evidence of lateral fields
extending onto the posterolateral parts of the shield
(Fig. 2A). The dorsal field is not visible on this speci-
men, and the medial limit of the lateral fields can be
traced (l.f., Fig. 2C) thanks to comparisons with the
second specimen (l.f., Figs. 1B2, 3).
One of the most remarkable features of Tannuaspis
is the ornamentation of the exoskeleton (Figs. 1, 2, 3),
which consists of closely set, pointed, and costulate
tubercles. Near the lateral and rostral margins of the
shield, these tubercles are strongly prominent, giving
the whole shield a serrated contour. On the ventral
surface of the abdominal division of the shield, as
well as on the ventral side of the marginal and rostral
regions, the tubercles are blunt and probably worn.
Measurements :
Total length of the shield (estimated) : 72 mm.
Maximum breadth : 64 mm.
Interorbital distance : 7 mm.
Preorbital distance : 21 mm.
~- THE SECOND SPECIMEN (PLAN 1013/2).
This specimen differs from the holotype by its
slightly larger size, greater preorbital distance and
possibly more regularly rounded anterior margin
(Figs. 1B,3). It consists of the dorsal exoskeleton and
its natural imprint, preserved in ventral, or internal
aspect. The exoskeleton is preserved on the left half of
the specimen. There, it shows a relatively smooth
basal layer, with some faint traces of subaponeurotic
blood vascular grooves. Impregnation with alcohol
yields a polygonal pattern which does not seem to
coincide with the limits of the individual tubercles.
This pattern is due to the presence of tesserae, bearing
several tubercles.
The median dorsal and the right lateral fields are
clearly visible (mdf., I.f., Fig. 1B2). Near the poste-
rior and lateral margins of the lateral field, a few iso-
lated tesserae remain attached to the surrounding
exoskeleton (tes., Fig. 1B2). These tesserae are relati-
vely small and seem to bear one or two tubercles. A
tentative reconstruction of the dermal covering of the
right lateral field is proposed in Figure 3.
The orbits are not clearly visible, and only that of
the right side (orb., Fig. 1B2) can be located. The
pineal region is very well marked by regularly arran-
ged tubercles (Fig. 4B), surrounding a smooth area
which may be a pineal foramen (pi.f., Fig. 4B),
although the absence of preserved exoskeleton in this
 -- 498 --

1...nhyp •

:,~,- rb.

O 0"-'-- : \
\ \
l I I.f.

<2 oc.r.

....... =p.lim.end.

Fig. 2 - Tannuaspis levenkoi OBRUCHEV. Attempted reconstructions of the shield based on the holotype (PIAN n ° 1013/1). A, from Obru-
chev (1956, fig. 2), dorsal view ; B, dorsal view (the position of the naso-hypophysial opening is hypothetical) ; C, ventral view.
Scale : 1 cm.
Reconstitutions du bouclier, bas6es sur l'holotype (PIAN n ° 1013/1). A, d'apr6s Obruchev (1956, fig. 2), vue dorsale ; B, vue dor-
sale (la position de l'orifice naso-hypophysaire est hypoth6tique) ; C, vue ventrale. Echelle : 1 cm.
 -- 499 --

,.,.--.;.,.. . . . .

!i7." .', . . . . "

• : . - " • . -..

~.?::.. : ....~ i!: i:-ii

-~-.::.-,.
• -.~..,,. :-.~

!
1 I
l I
l !
l I
l I
l !
!
/e

I I

Fig. 3 - - T a n n u a p i s cf. l e v e n k o i OBRUCHEV,a t t e m p t e d r e c o n s t r u c t i o n o f the shield, b a s e d on P I A N n ° 1013/2. The tesserae c o v e r i n g the

lateral field are r e c o n s t r u c t e d o n the r i g h t side only. Scale : 1 cm.
R e c o n s t i t t t t i o n bas6e sur le sp6cimen P I A N n ° 1013/2. Les tess~res c o u v r a n t les c h a m p s l a t 6 r a u x ne sont reconstitu6es que du c6t6
droit.

region makes it difficult to ascertain. In the recons-
truction presented here, we assume that an open
pineal foramen existed in Tannuaspis (Figs. 3, 4B), a
character which is widely distributed among osteos-
tracans [with only two exceptions : Acrotomaspis and
Gustavaspis (W~ngsj6, 1952 ; Janvier, 1981a)]. There
is no evidence of an independent pineal plate.
,/cr. B ..... ..::i;;~. Pi.f.
..: -~';',,:..::
~n ..~i~.::!~:::.~:!:~&
t I
or.n.
C tub,,
".::,~./_' ~ - , ~ .'-.'' _ ~ ~,
I I "I / less vertebrates known exclusively from the Lower
Fig. 4 - A,B, Tannuaspis cf. levenkoi OBRUCnEV, PlAN n °
1013/2 ; A, naso-hypophysial opening and crack (obli-
quely hatched), surface covered with exoskeleton dot-
ted, median axis of the shield indicated by a dashed
line ; B, pineal region (camera lucida drawing of the
cast of the specimen). C, Tuvaspis margaritae OnRU-
CrIEV, PIAN n ° 1013/7. Interpretation of the anterior
margin of the specimen. Scale : 1 mm.
A, orifice naso-hypophysaire et cassure (hachures obli-
ques), surface couverte par l'exosquelette en gris6, axe
m6dian du bouclier indiqu6 en pointill6; B, r6gion
pin6ale (dessin & la chambre claire du moulage du sp6ci-
men). C, interpr6tation de la pattie ant6rieure du sp6ci-
men. t~chelle : 1 mm.
The q u e s t i o n of the position of the naso-
hypophysial opening may be partly solved thank to
this specimen. Obruchev (1956, fig. 2 ; 1964, fig. 30)
figured a relatively small, rounded naso-hypophysial
opening in Tannuaspis. It was supposed to be situated
relatively far in front of the orbitopineal region. In
fact, the only possible evidence of such an opening is
a small circular area situated approximately 12 mm in
front of the pineal foramen (n.hyp., Fig. 4A). Unfor-
tunately, a large, curved crack passes through this
presumed opening (cr., Fig. 4A). Anyhow, if this
rounded area devoid of exoskeletal impression really
is the filling of the naso-hypophysial opening, it is
noteworthy that it is not surrounded by any marked
-- 500 --
naso-hypophysial depression and is situated very far
in front of the orbitopineal region, farther than in any
other known osteostracan. Although this interpreta-
tion of the naso-hypophysial opening still remains
uncertain, the only certainty we can have on this pro-
blem is that there is no evidence of a naso-
hypophysial foramen in the first twelve millimetres of
the prepineal part of the shield.
Finally, there is a median dorsal crest on the abdo-
minal division of the shield (Figs. 1 B2, 3), behind the
:.'.:);..:--,,~Z,~:.~:.:.~'~.:!,:(~::~ median dorsal field. The posterior limit of the shield
is not known in this specimen.
"k DISCUSSION
md.f. There is no doubt that Tannuapis belongs to the
Osteostraci, as evidenced by the lateral and median
dorsal fields, a feature found in no other craniate
group. Moreover, the structure of the exoskeleton of
I Tannuaspis, as far as it can be studied by immersion
in alcohol, is of osteostracan type, with vague indica-
tions of a tessellate pattern. Noteworthy is the striking
resemblance in the type of exoskeletal ornamentation
and even in the outline of the shield, between Tan-
~.. nuaspis and some of the Galeaspida, a group of jaw-
and Middle Devonian of China (e.g. Polybranchias-
pis, Liu, 1965, 1975 ; Hanyangaspis, P ' a n & Wang,
1975). This resemblance is, however, superficial and
is not indicative of close phylogenetic relationships.
The tubercles of galeaspids are hollowed by a large
<<pulp cavity >> and may be regarded, at least in the
primitive forms of the group, as individual dermal
units. In contrast, those of Tannuaspis are massive,
like those of other osteostracans.
The oralobranchial cavity of Tannuaspis, although
incompletely known, is of typical osteostracan type,
with large branchial fossae. It cannot be decided whe-
ther it is of orthobranchiale or oligobranchiate type.
The orbits are comparatively small, and more
widely spaced than in any other known osteostracan.
There is no pineal plate, a condition which may be
regarded as derived, since an independant pineal plate
is present in all primitive osteostracans (Janvier,
1981a, b, 1984, 1985). The loss of the pineal plate
occurs in several osteostracan groups, e.g. Mimetas-
pis, Didymaspis, kiaeraspidians and benneviaspi-
dians.
The naso-hypophysial opening, if circular in shape
as suggested by the only available indications, would
be similar to that of the Tremataspididae, in particu-
lar Tremataspis mammillata P A T T E N , where the
 - - 501 - -
opening itself is very short and situated at the bottom
of a small circular depression, corresponding to the
naso-hypophysial depression of other osteostracans.
The posterolateral extension of the endoskeleton,
along the ventrolateral margins of the abdominal divi-
sion of the shield (v.l.l., Figs. 1A1, 2) is similar to that
met with in the Tremataspididae (Denison, 1951, fig.
22) and in Didymaspis (P.J. pers. observ.). However,
this endoskeletal lamina extends posteriorly farther
than in any known Tremataspididae.
Tannuaspis differs from all the other known osteos-
, tracans by the apparent absence of lateral sensory-line
grooves, by the type of ornamentation, and by its very
short lateral fields. In fact, when the lateral fields
tend to become reduced in osteostracans, they often
are subdivided into two parts (Tremataspis, Oeselas-
pis) or more (Acrotomaspididae, Nectaspidinae),
whereas in Tannuaspis, they remain single, but are
very short, and situated at the same level as the
median dorsal field. This type of lateral fields is sug-
gestive of Thyestes. The number of s.e.1, canals lea-
ding to the lateral fields is not known ; only a faint
trace of such a canal is seen on the holotype (s.e.l.,
Fig. 1A1).
The affinities of Tannuaspis are still obscure, yet
one may propose hypotheses of relationships on the
basis of the data presented here. Obruchev (1964)
considered that the family Tannuaspididae had to be
included in the Tremataspidida, whereas Tarlo (1967)
gave the Tannuaspidida the same rank (suborder) as
the Tremataspidida, Sclerodontida and Didymaspi-
dida, whithin the order Tremataspidiformes. Other
authors, such as Stensi6 (1964) and Moy-Thomas &
Miles (1971) did not assign Tannuaspis to any particu-
lar systematic position.
Considering the view held by Westoll (1945), Deni-
son (1951) and Halstead (1982), that the Tremataspi-
didae are primitive osteostracans, and that paired fins
have been acquired several times in the more advan-
ced forms of the group, Tannuaspis, being devoid of
paired fins, would appear as a primitive osteostracan.
However, this hypothesis of osteostracan phylogeny
has been criticized (StensiO, 1927, 1964;Heintz,
1939 ; W~ngsjO, 1952 ; Janvier, 1981a, 1984, 1985),
because many characters are shared only by the Tre-
mataspididae and some particular cornuate osteostra-
cans (Procephalaspis, Auchenaspis, Thyestes) : the
course of the infraorbital sensory-line, the shape of
the pineal plate, the histologic structure of the tuber-
cles of the exoskeleton, the toothed supraoral field
(with Thyestes only), and some details of the internal
anatomy. According to this view, the Tremataspidi-
dae would not be primitive osteostracans, but a highly
derived group of cornuate osteostracans, closely
allied to Thyestes, and which would have lost their
paired fins secondarily. In this respect, Tannuaspis,
which is devoid of paired fins, and shares with the
Tremataspididae a short naso-hypophysial opening
and small lateral and median fields, would have to be
regarded as a much derived osteostracan. However,
Tannuaspis differs from the Tremataspididae by the
lack of pineal plate, the widely separate orbits, the
type of ornamentation (yet costulate tubercles also
occur in primitive thyestidians, such as Procephalas-
pis and Auchenaspis), and the broad occipital region
of the endoskeleton.
3.1.2. Genus Tuvaspis OBRUCHEV,1956
Tuvaspis margaritae OBRUCHEV,1956
(Fig. 4C)
Tuvaspis margaritae (Obruchev, 1956, 1961, 1964)
is a poorly known species, also from Tuva, and repre-
sented by the holotype (PIAN n ° 1013/7) and nume-
rous scattered scales. The exoskeleton is covered with
pointed and costulate tubercles which are similar to
those of Tannuaspis. The anterior margin of the
holotype (Fig. 4C) shows an internal notch covered
with perichondral bone and which may be interpreted
as the oral notch (or.m, Fig. 4C). Nothing more can
be said on this species which, according to its type of
ornamentation, is probably very close to Tannuaspis
and should probably be placed with the latter in the
Tannuaspididae.
3.2. Genus Ilemoraspis OBRUCHEV,1961
Iiemoraspis kirkinskayae OBRUCHEV,1961
(Figs. 5-7)
Ilemoraspis kirkinskayae (Obruchev, 1961, 1964) is
only known from the holotype (PIAN n ° 1642/1),
which consists of a crushed cephalic shield and a part
of the trunk scales. No reconstruction has ever been
proposed for this species and we give here an attemp-
ted reconstruction of the shield (Fig. 6), based on a
detailed analysis of the hototype (Fig. 5). The distor-
tion of the shield made this reconstruction difficult to
realise, many of the tesserae being displaced and over-
lapping each other.
The exoskeleton of the cephalic shield of L kirkins-
kayae consists of large, polygonal tesserae, most of
 -- 502 --

n.hyp, x.

..orb.
pi.f.. ~
~.,~ ~ ~ , ,
~~+:~ ,~
m d.f..L ~ ' ~ , ' ~ ~ ~ , , \ \
'~~'Ti'it -~-- I.f.
,%
....x,//

"x .~'<"-' +'" ~ "" "'+

"-" T,/.~..,.,<,,,! ,,,,~~~:4,<~,, ....
"I / L~: ' k' ~ "I I,,¢,,,.,,
' , , . ,I+tttJ,
: ' o, tr • ~ ' . . ~ . r ~ , ,," ~ l
~. \ ~.~,r/

iI,~ "
m.d.cr.

SC.
/
/
/

~ LL_ ¸¸¸7¸0 J i. . . . . . . . .

Fig. 5 - - Ilemoraspiskirkinskayae OBRUCnEV, Middle Devonian, Khakassia, USSR, Holotype, P I A N n ° 1642/1. A, rubber cast of the speci-
men, dorsal view ; B, interpretation of the specimen ; C, SEM photograph of the ornamentation of the tesserae covering the lateral
fields. Scale : 1 cm (A,B) and 1 m m (C).
A, moulage en 61astom6re du sp6cimen, vue dorsale ; B, interpr&ation du sp6cimen ; C, ornementation des tess6res couvrant les
champs lat6raux (photo. MEB). ]~chelle : I cm (A,B) et 1 m m (C).
 - - 503 - -

them being preserved only as impressions. However,
small fragments of the exoskeleton are still preserved
and clearly show the radiating canals of the middle
layer, as in typical osteostracans. The ornamentation
consists of thin ridges forming bundles, but almost
never radiating from the centre of the tesserae. In
contrast, the bundles of ridges follow the limits of
exoskeletal discontinuities, such as the lateral fields,
orbits, naso-hypophysial opening and lateral-line
grooves. Although much distorted, the orbits are, as
reconstructed here, relatively large and widely separa-
ted. The pineal region is indicated by changes in the
orientation of the ornamentation, on a small, median
tessera (pi.f., Fig. 6), but there is no clear evidence of
a pineal foramen proper. There is no independent
pineal plate.

n.hyp~
pif. \ iorb. x.
•, \ I /

I m.d .f.

Fig. 6 - - llemoraspiskirkinskayae OBRUCHEV, a t t e m p t e d r e c o n s t r u c t i o n of the shield, based on the h o l o t y p e ; d o r s a l view. Scale : 1 cm.
R e c o n s t i t u t i o n d u bouclier, bas6e sur l ' h o l o t y p e ; vue dorsale, l~chelle : 1 cm.

The naso-hypophysial opening seems to be small
and rounded in shape (n.hyp., Fig. 6). Its anterior
margin is slightly embayed by a narrow groove, which
is assymetrical in position (turning to the right), and
marked by a break in the course of the ridges of the
ornamentation.
The lateral and median fields are small and have
nearly the same size and position as those of Tannuas-
pis (l.f., m.d.f., Fig. 6). They are covered with small
tesserae, each of them bearing several short ridges
(Fig. 5C).
The sensory-line grooves are indicated by the distur-
bance they cause in the orientation of the ridges of the
ornamentation. In most cases, the grooves are per-
pendicular to the ridges, which become slightly flexed
at the point they meet the grooves (Fig. 6). One can
recognize a clear transverse line (transv., Fig. 6), and
an infraorbital line (iorb., Fig. 6).
 -- 504-
The most peculiar feature of the shield of L kirkins-
kayae is an elongated gap (x., Fig. 6), situated antero-
laterally to the orbit. At first glance, it resembles a
simple gap between disjointed tesserae, as it occurs
elsewhere in the specimen. But a careful examination
reveals that the margin of this gap are rounded in sec-
tion (Fig. 7B), and that the ridges of the ornamenta-
tion follow them regularly, as they do around the late-
ral fields and the orbits. This gap can hardly be an
anterior division of the lateral field (assuming that the
latter was divided into two parts, like that of Trema-
taspis) : it does not show any covering tesserae and
has a wrong orientation. It may also correspond to
the passage of a sensory-line groove, but it does not
look like that of other sensory-lines which, as mentio-
ned above, are perpendicular - and not parallel - to
the ridges of the ornementation. Morever, if the latter
interpretation is correct, one has to admit that a sup-
plementary sensory-line, parallel to the infraorbital
line, existed in this species. In sum the interpretation
of this peculiar gap is not yet clear and requires addi-
tional material.
f JSff \ \t
Fig. 7 - - Ilemoraspis kirkinskayae OnnUCHEV, camera lucida dra-
wing of the tesserae surrounding the slit-like (x) opening
in the anterolateral part of the shield. A, dorsal view ;
B, transverse section showing the rounded edges of the
opening. Scale : 1 m m .
Dessin /1 la chambre claire des tess6res entourant
l'ouverture allong6e (x) de la partie ant6ro-lat6rale du
bouclier. A, vue dorsale ; B, section transversale mon-
trant les bords arrondis de cette ouverture, l~chelle : 1
mill.
The shield probably possessed a median dorsal crest
(m.d.er., Fig. 5B) on the abdominal division. This
part of the shield is remarkably short, yet distorted
and difficult to examine on this specimen in its pre-
sent state of preservation. The posterior limit of the
shield can be observed on the right side but, as it is
often the case in thyestidians (e.g., Thyestes, Dartmu-
thia) there is a gradual passage between the abdomi-
nal division of the shield and the trunk scales, the for-
mer showing traces of scale rows incorporated to the
shield.
The marginal region of the shield is covered with
curved tesserae, like in most other osteostracans, and
their ornamentation is oriented parallely to the mar-
gin of the shield. Traces of these marginal tesserae can
be followed backwards further than the level of the
posterior limit of the shield (Fig. 6), but whether these
tesserae cover the lateral margin of a cornual process
or that of an elongated abdominal division is still
undecided. Neither is there any evidence of paired
fins, despite the presence of small scattered scales on
the left side of the specimen, and no pectoral sinus
can be observed.
The scales covering the trunk are broad, elongated
in shape, and ornamented with longitudinal ridges,
which are thinner than those of the exoskeleton of the
shield. About 1 cm behind the posterior limit of the
shield , some of the flange scales bear a prominent
cusp, which, when aligned, forms a serrated longitu-
dinal ridge. At least two such ridges occured on the
dorsolateral surface of the trunk.
~k D I S C U S S I O N
Ilemoraspis kirkinskayae is undoubtebly an osteos-
tracan as evidenced by its lateral and dorsal fields.
However, it differs from most other representatives
of the group, by its widely spaced orbits and unusual
type of ornamentation. As to the aspect of its exoske-
leton, it bears some similarities with (( Cephalaspis >>
gabrielsei DINELEY & LOEFFLER, by the presence of
large and lose tesserae with. an oriented ornamenta-
tion. However, the most striking resemblance is with
Tannuaspis, despite differences in the ornamentation
and shape of the shield. Ilemoraspis and Tannuaspis
share the following characters :
1) the widely separated orbits, 2) the lack of a pineal
plate, 3) the shape and the position of the lateral
fields. These resemblances, however, deserve some
comments ."the lack of pineal plate also occur in other
osteostracans (kiaeraspidians, benneviaspidians,
Mimetaspis), but in no other known osteostracan are
the orbits as widely separated as in these two taxa.
 - - 505 - -
The shape and the position of the lateral fields of
Tannuaspis and Ilemoraspis is not met with in other
osteostracans, even in forms with short lateral fields
such as Procephalaspis, Thyestes and the Tremataspi-
didae. Moreover, the naso-hypophysial opening
seems to have nearly the same shape in llemoraspis
4. C O N C L U S I O N S
When Tannuaspis is considered alone, it may be
placed among the Tremataspididae, and, thus, not in
a family of its own (the Tannuaspididae) but, at most
in a subfamily , the Tannuaspidinae, including also
Tuvaspis. This position is implied by the absence of
paired fins, the elongated abdominal division and, to
some extent, the short lateral and median dorsal
fields. It may be regarded as the sister-group of the
Tremataspidinae (Tremataspis and Oeselaspis) with
which it shares the short and rounded naso-
hypophysial opening.
REFERENCES
DENISON R.H. (1951) - Evolution and classification of the
Osteostraci. Fieldiana : Geology, Chicago, 11, 3, 157-
193.
HALSTEAD L.B.H. (1982) - Evolutionary trends and the
phylogeny of the Agnatha in K.A. JOYSEY & A.E. FRI-
DAY eds : Problems of Phylogenetic Reconstruction,
Academic Press, London, 159-196.
HEINTZ A. (1939) - Cephalaspida from Downtonian of
Norway. Skr. norske Vidensk. Akad (Mat. Naturv. KI.),
Oslo, (5), 5-110.
JANVIER P. (1981a) - Norselaspis glacialis n.g., n.sp., et les
relations phylog6n6tiques entre les Kiaeraspidiens
(Osteostraci) du D6vonien inf6rieur du Spitsberg.
Palaeovertebrata, Montpellier, 11 (2-3), 19-131.
JANVIER P. (1981b) - The phylogeny of the Craniata, with
particular reference to the significance of fossil <<agna-
thans ,. Jr. Vert. Paleont., Norman, 1 (2), 121-15n.
JANVIER P. (1984) - Preliminary description of Lower
Devonian Osteostraci from Podolia (Ukrainian SSR).
Bull. Br. Mus. nat. Hist. (GeoL), London, 38 (4), 309-
334.
and Tannuaspis. Thus, despite profound differences
in overall morphology, we would be inclined toward
considering Ilemoraspis as more closely related to
Tannuaspis (and Tuvaspis) than to any other osteos-
tracan.
However, when considering the hypothesis that
Tannuaspis, Tuvaspis a n d Ilemoraspis form a
monophyletic group characterized by the widely sepa-
rated orbits and short lateral and median dorsal
fields, a position of this group within the Tremataspi-
didae becomes unlikely, since llemoraspis has a short
shield. In this latter case, the resemblance between
Tannuaspis and the Tremataspidinae would be due to
parallelism. This latter hypothesis would be more
consistent with the geography, since Tannuaspis,
Tuvaspis and Ilemoraspis have hitherto been found
only in this region of Central Asia.
JANVIER P. (1985) - Les C6phalaspides du Spitsberg.
Anatomie, phylog6nie et syst6matique des Ost6ostrac6s
siluro-d6voniens. R&ision des Ost6ostrac6s de la Forma-
tion de Wood Bay (D6vonien inf6rieur du Spitsberg).
Cahiers Pal~ont., Paris, 220 p.
LIU Y.H. (1965) - New Devonian agnathans from Yunnan.
Vert. Palasiat., Peking, 11 (2), 132-143 (in Chinese with
English summary).
LIU Y.H. (1975) - Lower Devonian agnathans from Yunnan
and Sichuan. Vert. Palasiat, Peking, 13 (4), 202-216 (in
Chinese with English summary).
MELESTSHENKO V.S., PREDTECHENSKIY N.N. &
YANOV E.N. (1973) - Intramontane depressions of the
Altai-Sayansk folded region in Stratigraphy of the
USSR, Devonian system, Nedra, Moscow, 2, 95-118 (in
Russian).
MOY-THOMAS J.A. & MILES R.S. (1971) - Palaeozoic
Fishes, Chapman & Hall, London, 259 p.
OBRUCHEV D.V. (1956) - Cephalaspids from the Lower
Devonian of Tuva. Dokhl. Akad. Nauk. SSSR, Mos-
cow, 106 (5), 917-919.
 - - 506 - -

OBRUCHEV D.V. (1961) - Class Ostracodermi in Biostrati-
graphie du Pal6ozoYque des monts Sa~an-Alta~, Trudy
SNIlGGIMS (Siberian Research Institute of Geology,
Geophysics and Mineral Resources), Moscow, 20, 560-
561 (in Russian).
OBRUCHEV D.V. (1964) - Class Osteostraci in I.A. ORLOV
ed. • Fundamentals of Palaeontology, Nauka, Moscow,
11, 84-107 (in Russian).
P'AN K., WANG S.T. & LIU Y.P. (1975) - The Lower Devo-
nian Agnatha and Pisces from South China in Professio-
nal Papers in Stratigraphy and Geology, Geological
Press, Peking, 1, 135-301 (in Chinese).
PREDTECHENSKIY N.N. (1975) - The Kendeyskaya Group
and Samagaltayskaya Group in Stratigraphic dictionary
of the USSR, Cambrian, Ordovician, Silurian, Devo-
nian, Nedra, Leningrad, p. 222, 382 (in Russian).
RZONSNITSKAYA M.A. & MAYMINA L.G. (1975) - The
Ilemorovskaya Group in Stratigraphic dictionary of the
USSR, Cambrian, Ordovician, Silurian, Devonian,
Nedra, Leningrad, p. 179 (in Russian).
STENSIO E.A. (1927) - The Downtonian and Devonian
vertebrates of Spitsbergen. 1. Family Cephalaspidae.
Skr. Svalbard Ishavet, Oslo, 12, 1-391.
STENSIO E.A. (1964) - Les Cyclostomes fossiles ou Ostra-
codermes, in J. Piveteau : Trait6 de Pal6ontologie, Mas-
son ddit., Paris, 4 (1), 173-425.
TARLO L.B. & HALSTEAD (1967) - Agnatha in W.B. HAR-
LAND & alii eds, London Geological Society, London,
629-636.
VLADIMIRSKAYA E.V. (1975) - The Chondergeyskaya
Group in Stratigraphic dictionary of the USSR, Cam-
brian, Ordovician, Silurian, Devonian, Nedra, Lenin-
grad, p. 489.
WANGSJO G. (1952) - The Downtonian and Devonian
vertebrates of Spitsbergen. 9. Morphologic and systema-
tic studies of the Spitsbergen cephalaspids. Norsk Pola-
rinst. Skr., Oslo, 97, 1-611.
WESTOLL T.S. (1945) - A new cephalaspid fish from the
Devonian of Scotland, with notes on the structure and
classification of ostracoderms. Trans. Roy. Soc. Edin-
burgh, Edinburgh, 61, 341-357.

M a n u s c r i t d6finitif regu le 17.01.1985