Biological Journal of the Linnean Society, 2002, 75, 367-377. With 4 figures
Intraspecific variation in limblessness in vertebrates: a
unique example of microevolution
DOMINIQUE ADRIAENSl", STIJN DEVAEREl, GUY G. TEUGELS',
BARBARA DEKEGEL' and WALTER VERRAES'
'Ghent University, Institute of Zoology, K.L. Ledeganckstraat 35, B-9000, Gent, Belgium
2Africamuseum, Department of Zoology, B-3080 Teruuren, and KULeuven, Department of
Biology, B-3000, Leuven, Belgium
Received 5 May 2001; accepted for publication 13 November 2001
The variation in limb reduction in vertebrates has generally been observed as characterizing higher-level taxa. Such
structural changes are thus considered to reflect macroevolutionary processes. A statistical analysis of metric variables of some species of the African catfish family, Clariidae, suggests that fin reduction occurs a t the microevolutionary level as well. In a t least three species of that family intraspecific variation in the presence/absence of the
pelvic fins was observed, and one species also showed similar variation for the pectoral fins. Discriminant function
analysis confirmed that the variation is intraspecific, and even occurs within the same population. Sexual dimorphism could be excluded. This variation can be observed in the most anguilliform species of the clariid family, suggesting a link with body elongation (as is the case in tetrapods that show limb reduction). Pelvic fin loss appears
to precede pectoral fin reduction during evolution. From the morphology it could be ascertained that the loss of the
pelvic fins is coupled to the loss of the pelvic girdle, contrary to the case for the pectoral fins and girdle. Differences
in functionality may explain this. Breeding results support the occurrence of intraspecific variation, as an F1 offspring showed a difference compared with the parents. For at least one species, the benefit from body elongation
and limb reduction can be related to its highly specialized life style, as it lives subterraneanly in muddy soil of the
Central African rainforest. 02002 The Linnean Society of London, Biological Journal of the Linnean Society, 75,
2002, 367-377.
ADDITIONAL KEYWORDS: catfish
-
Clariidae - intraspecific variation - limblessness
INTRODUCTION
The developmental processes inducing differentiations
of limbs lie a t the base of one of the most important
evolutionary innovations, namely that variations
in overall limb morphology reflect changes on a
macroevolutionary level, i.e. they characterize higher
level taxa (Futuyma, 1998). One of the macroevolutionary traits involved with limb morphology is the
presence or absence of limbs. Limb reductions have
occurred independently in many higher level taxa
of vertebrate lineages, as for example teleosts (Bell,
1987; Nelson, 1989-90), amphibians (Duellman &
Trueb, 19861,reptiles (Gans, 1975; Motani et al., 1996;
Carroll, 1997; Pough et al., 1998), and even some
~
Torresponding author. E-mail: [email protected]
0 2002 The Linnean Society of London, Biological Journal
of the
-
microevolution.
mammals (Lande, 19781, all both in recent and fossil
forms. Some intermediate forms, which could be a
reflection of an evolutionary grade towards limb
reduction and body elongation, do exist up to the interspecific level (e.g. in the genus Lerista within the Scincidae) (Lande, 1978; Greer, 1987, 1990).As this is still
above the species level, it technically still involves
processes underlying macroevolution (Futuyma, 1998;
Carroll, 2001). As discussed by the latter authors, the
notion of a gap between processes underlying microand macroevolution ranges between being substantial
to nonexistent, depending on different methodological
approaches (i.e. palaeontology vs. genetics, respectively). As Carroll (2001) mentioned: 'the distinction
between macro- and microevolution in terms of
morphological change is descriptive, not mechanistic'.
Evidence is provided in this paper that even such a
trait as the presence/absence of limbs can vary intraLinnean Society, 2002, 75, 367-377
367
368
1). ADRIAENS ET
AL
specifically in certain vertebrates and thus reflects
both micro- and macroevolutionary processes.
In most vertebrate lineages characterized by limb
reduction to limb loss the body undergoes marked
elongation, which is considered a s being an adaptation
to burro\ving and fossorial life styles (Gans, 1973,
1975: Withers, 19811. In teleost fishes, many strategies of burrowing behaviour exist in elongated
species (Nelson. 1994). and one unique example is the
clariid catfish family, in which a complete range of
body forms exists from fusiform (with both pelvic and
pectoral fins well developed) to anguilliform (with
all paired appendages lacking) (Boulenger, 1908;
Pellegi.in, 1927i. For some species, it had been noted
that the pelvic and pectoral fins may be reduced or
sometimes absent iBoulenger, 1911; Poll, 1942).
I n t h i s paper, we have used statistical means t o
support the hypothesis of interspecific, but more
importantly intraspecific and even intrapopulation
variation i n the presencelabsence of paired fins in
several species of the Clariidae. Evidence from the
morphology suggests variation in an early triggering
of fin development, whereas breeding results suggest
a large degree of flexibility. Field observations support
the hypothesis that within a t least one of these species
limb reduction and body elongation is related to a
highly specialized fossorial habit, unique for teleost
fishes.
MATERIAL AND METHODS
MEASUREMENTS
Using digital callipers (0.05 mm accuracy), 30 metric
measurements were taken on 473 specimens belonging to nine clariid species: Clarias gariepinus
j Rurchcll. 1822 I, Clarias salae Hubrecht, 1881, Clarinllahc~s 1oiig.icauda (Boulenger, 1902, Gymnallabes
olt,ar.cm Roman, 1970, G. t-ypus Giinther, 1867,
Ptaf,vcI(~rla.srnacha,doi Poll, 1977, Platyallabes tihoni
(Poll. 19441, Cliannallahes apus (Gunther, 1873) and
Dolichallahr~s inicrophthalmus Poll 1942. Specimens
were obtained from the collections of the MRAC
1 Musee Royal d’Afrique Central, Belgium), BMNH
(British Museum of Natural History), MNHN (Musee
National d’Histoire Naturel, Paris), MHNG (Musee
d’Histoirc Naturelle. Geneve), NMW (Naturhishe.; Museum, Wieni, MCZ (Museum of Comparativit ZOOk~gY. Harvard), ZMB (Museum fur
Yaturkundt~,Berlin) and RMNH (Rijksmuseum van
Natuurl ijke Historie, Leideni. Measurements taken
ivcre: 1 I total length, (21
standard length, (3) preanal
length ( 4I pre-dorsal-fin length, ( 5 ) distance between
the dorsal fin and the supraoccipital process, ( 6 )
caudal peduncle depth, ( 7 ) abdominal depth, (8)maxillary harlwl length, 19) internal mandibular barbel
length, (10) external mandibular barbel length, (111
nasal barbel length, (12) skull length, ( 1 3 )skull width,
(14) skull height, ( 15) preorbital skull length, i16)
orbital skull width, (17) orbital skull height, ( 1 8 )
length of the supraoccipital spine, (19) width of the
supraoccipital spine, (20) interorbital distance, 121)
distance between the anterior nostrils, (22, distance
between the posterior nostrils, (23) orbital diameter,
(24) snout width, ( 2 5 ) snout height, (26) distance
between the snout and the hyoid fold, (27) distance
between the internal mandibular barbel bases, (28)
distance between the external mandibular barbel
bases, (29)mouth width, and (30)caudal fin length Ifor
a drawing of the taken measurements, we refer to
Teugels (1986il.
STATISTICAI, ANAI’YSIS
Backward stepwise Discriminant Function Analyses
(DFA) were performed using Statistica (Statsoft, Inc. ),
with the number of variables for each analysis depending on the amount of correlation between variables,
and correlation between means and variance (correlated variables and variables with a high correlation
(>0.9) between means and variance were removed
from the analysis). The first analysis involved all
species, which were categorized into 13 groups (based
on the configuration of the paired fins) as follows: ( 1 1
cg: Clarias gariepinus ( N = 21 1, 12) Cs: Clarias salae
( N = 411, (3) Cl: Clariallabes longicauda ( N = 41i, ( 4 )
Ga: Gyrnnallabes alvarezi ( N = 12j, 15) Ga-c: G.
alvarezi lacking pelvic fins ( N = 5), ( 6 )Gt: Gyiiznallabes
typus ( N =31), (7)Pm: Platyclarias rnachadoi ( N = 17),
(8)Pt: Platyallabes tihoni ( N = 53), ( 9 ) Ca: Channallubes apus ( N = 4), (10) Ca-v: C. apus lacking pelvic
fins ( N = 1551, (11)Ca-uc: C. apus lacking both pectoral and pelvic fins ( N = 76i, (12)Drn: Dolichallabes
microphthalmus ( N = 31, and (13)Dm-L>:D. inicrophthalmus lacking pelvic fins (N = 14). Of all these
groups, these not designated a s ‘-v’ or ‘-ix’thus possessed both pectoral and pelvic fins (in some rare cases
where specimens only had one pectoral or pelvic fin
these were grouped with those having both contralatera1 fins). For the analysis of the geographical variation in C. apus, five populations were used in which
specimens with and specimens without paired fins
could be observed: (1)BL: Boloho, western part of the
Middle Zaire river system (Likouala river, Congo
Brazzaville), (2) BM: Bumba, northern part of the
Middle Zaire river system (Congo Republic), (31 LD:
Loadjili, coastal river system (Congo Brazzaville 1, ( 4J
TM: Tchimenji, Kasai river system (Angola),and ( 5 )
WO: Woleu, Woleu-Ntem river system (Gabon). The
coding ‘-u’ and ‘-vc’ follows that as above. For the
analysis of possible sexual dimorphism, five groups
could be defined: (1)m, male specimens having both
-L‘ LOO2 The Linnean Society of London, Biological Journal of the Linnenri Society. 2002. 75, 367-377
369
INTRASPECIFIC VARIATION IN LIMBLESSNESS
pectoral and pelvic fins, (2) m-u, males lacking only
pelvic fins, (3) m-uc, males lacking both pectoral and
pelvic fins, (4)f u , females lacking only pelvic fins, and
( 5 )f u c , females lacking both pectoral and pelvic fins.
Only one female specimen with both pectoral and
pelvic fins present was available. It was thus not
included in this analysis.
Discriminant Function Analysis
MORPHOLOGY
The morphology of the pectoral and pelvic fins and
girdle was studied using cleared and stained material.
Clearing and staining followed Taylor & Van Dyke
(1985). Specimens used for the clearing came from
a single sample of the MRAC (KMMA 88-5-P2192-227). Breeding results come from spontaneous
reproduction of two individuals that were commercially obtained.
-10
b
Pelvics absent. pectorals present
\
\
RESULTS
STATISTICAL ANALYSIS
A first discriminant function analysis was undertaken
to determine to what degree the metric variables could
be used to reflect interspecific variation, as well as
to evaluate the relative position of the intraspecific
groups (with specimens having a different number of
paired appendages). A plot of the 13 groups ( N = 473),
based on 30 variables reflects a cluster structure, supported by what is to be expected from the general morphology and taxonomy: all species (not groups) differ
significantly from each other ( P < 0.001 for all pairwise contrasts between species, with the exception of
P = 0.003 for Ga-ulCa, Fig. la). The squared Mahalanobis distances (D'-values) between species ranged
between 14 (for GtlCa) and 542 (for CglDm, Table 1).
Root 1reflects the degree of anguilliformity (supported
by the factor structure). The fusiform Clarias gariepinus has the lowest root 1 values, whilst Clarias salae
lies close to the slightly anguilliform Clariallabes
longicauda, which is to be expected because C. salae
also has a slightly elongated body (reflected in the
taxonomic position in the subgenus Anguilloclarias)
(Teugels, 1986).A chain of consecutive clusters along
root 1 includes (starting with the less anguilliform):
Platyclarias machadoi, Platyallabes tihoni, Gymnallabes aluarezi, G. typus, Channallabes apus and finally
the extremely elongated Dolichallabes microphthalmus. The species P. tihoni is separated from this
cluster along root 2 as a result of the anterior-placed
dorsal fin. Of importance is the relation between the
intraspecific groups: all groups with a different
number of paired fins cluster together according to
the species they belong to. The lowest D"-values were
observed between the groups of C. apus and those of
D. microphthalmus, where none were significantly
02002 The Linnean Society of London, Biological Journal
of
I=
4
. ' 1
I
2
3
4
w
5
i
Root 1
Figure 1. Discriminant Function Analysis (DFA) of nine
clariid species, showing observed ranges of scores (polygons) and centroids (crosses) along root 1 vs. root 2. Solid
lines represent specimens with both pectoral and pelvic fins
present, dashed lines represent specimens lacking only the
pelvic fins, dotted lines represent specimens lacking both
pelvic and pectoral fins. (a) Total data matrix (N= 473) of
13 groups (belonging t o nine different species), based on
15 metric characters. (b) Channallabes apus specimens
(N= 2531, based on 30 metric characters. (c) Five geographical groups of C. upus (N= 1141, based on 30 metric
characters.
different from one other: 1.7 between Ca and Ca-u
( P = 0.899), 5.6 between Ca and Ca-uc ( P = 0.107), 8.1
between D m and Dm-u ( P = 0.192). Although the latter
groups ( D m and Dm-u) lie separate on the plot of root
1vs. root 2, this is not the case for plots of other roots.
The groups Ca-u and Ca-uc, a s well as the groups
Ga and Ga-u, were significantly different ( P <
the Linnenn. Society, 2002, 75, 367-377
370
D. ADRIAENS ET AL.
Table 1. Values of f’ (upper right) and Squared Mahalanobis distances (D’, lower left) of all 13 groups, reflecting the
significant discrimination between the species, but not between groups within a single species
C1
0 899
Cs
0.107
2
6
2
,328
118
125
341
13ti
144
35
40
5.3
23
21
11.5
121
41
49
43
16
105
115
f’
0 01
I’
<
385
166
173
31
33
74
35
30
141
136
Ca - u
0.447
C’tr
..,..,.
:.
......
......
.._.
......
_:.
_..
148
92
511
542
236
285
230
223
266
<. ...
......
......
.,..,.
16
235
272
55
73
85
64
127
261
294
68
84
74
92
139
1
6
Ca-cc
......
....,.
2
~
P c 0 01.
+.,.
.._..
Ga
Gn-c
Gt
Prrt
91
87
106
Pt
_._
......
......
.;._..
.,,.,.
.,..,.
......
.,..,.
0.192
8
108
83
74
187
132
136
101
86
229
180
19
31
77
73
17
83
102
0 001:D’ < 20 underlined.
CO
C’o-l~c~
Dni-c.
......
Table 2. Values of P (upper right) and Squared MahaInnobis distances iD’, lower left) of the Channallabes apus
groups
(“I-L
Dm
P < 0 001
0.001,. however, the D’-value was very low (2.3 and
19, respectively 1. Comparable low D’-values could be
observed between some groups of different species:
C k - r and Ca i 16), Gt and Ca (14),Cs and C1 (161, and
Cra-15 and Gt ( 1 7 ) (all P < 0.001, except for Ga-u and
Ca with P = 0.003).
The possibility t h a t true intraspecific variation
might be masked by the total interspecific variation
could be ruled out, based on a DFA of only the Chantiallahes a p u s specimens (Fig. l b ) . True intraspecific
variat.ion is again suggested because the distribution
of the c‘. aprrs groups is comparable to that of the total
data set ( w i t h the exception of a n inverted root 1).
Again. D’-values between the groups remain very low:
0.9 iP =- 0.447, between Ca and Ca-u; 5.8 ( P = 0.001)
between Ca and Ca-uc; and 2.2 ( P < 0.0011 between
Ca-u and Ca-oc (Table 21. The seemingly significant
difference between Ca and the other groups has to be
viewed with caution as Ca contained only four specimens, this in contrast to the much higher number for
the other groups. It remains clear that the distance
between the groups is very low, as is reflected in the
D2-values and the large overlap of the polygons on the
DFA.
Other possible factors underlying the variability in
fin development, as for example geographical variation between populations or sexual dimorphism, could
be ruled out as well. A DFA of five geographically
separated populations (not only regionally, but also
by river system), suggests t h a t even within a single
population the number of paired fins may vary
(Fig. lc). As could be expected, some populations
cluster more closely together than others, but it is the
relation within the populations that is important.
Within four of the populations the groups with a different number of paired appendages could not be distinguished ( P > 0.1, Table 3 ) and cluster closely
together: Boloko population (D’ = 0.71, Bumba population (D‘ = 2.41, Loadjili population (D’ = 1.41, and
Tchimenji population (D’= 0.5). Only the Woleu
population showed two distinct groups tP < 0.001,
D2 = 20.71, largely based on a narrower supraoccipital
width. However, the results for this population should
be treated with caution, as the group with taxa having
all paired fins contained only two specimens.
Analysis of the two sexes tN = 149) showed no indication of a strict sexual dimorphism in paired fin
development (Table 4).High P values (>0.01)indicated
a non-significant distinction between sexes for m and
f - u , for m and f-uc, for m-u and f - v , for m - L r and f-uc.
For some of them, even the D2-values were extremely
low (e.g. 0.018 between f-uc and m-uc).
BREEDINGRESULTS
The hypothesis of true intraspecific variation within
Channallabes apus was confirmed by the production
0 2002 The Linnean Society of London, Biological Journal
of
the Linriean Society, 2002, 75. 367-377
INTRASPECIFIC VARIATION IN LIMBLESSNESS
37 1
Table 3. Values of P (upper right) and Squared Mahalanobis distances (D’, lower left) of the geographical populations of
Channallabes apus groups
BL-u
BL-u
BL-vc
BM
BM-u
LD-L1
LD-uc
TM-u
TM-uc
wo
wo-u
BL-vc
0.650
1
17
9
1
3
2
4
9
18
BM
BM-u
*:3
::::i:
:i:
20
9
2
2
0
2
7
12
*
LD-u
0.456
0.325
:::*
*:*
*$
0.109
2
17
19
10
15
32
18
27
32
22
29
49
33
TM-u
0.057
0.265
0.432
0.867
**
* ..,
0.337
1
4
6
5
24
2
2
2
17
0.254
0.399
TM-uc
*
0.276
0.041
0.115
**:
.....,.
:. .:.
.._
.>
.!..:_
.,..,.
0.012
0.014
0.809
0
6
8
WO
0.188
0.457
0.211
0.127
5
7
WO-r:
_.....
.:. . .
_,._.
..~..,
.,. .
...
.._
.
.:.:::,
...
_._
0.023
:;:.,,
.,....
..
21
< 0.01; :“:‘P< 0.001; o-’ 5 5 underlined.
‘I“
Table 4. Values of P (upper right) and Squared Mahalanobis distances (D2,lower left) of sex groups of Channallabes apus groups
m
m
m-u
m-uc
f-u
f-uc
.’.
LD-vc
m-u
m-uc
f-u
f-uc
0.714
0.069
0.932
0.018
0.064
:i
0.3
3.0
0.1
3.3
*
2.3
0.3
2.5
2.1
0.0
**
**
0.932
**
2.4
P < 0.001; D2 < 1 underlined.
of offspring. Two surviving larvae of spontaneous
reproduction in an experimental aquarium exhibited
a different paired fin morphology from that of their
parents. Although both parents lacked both pectoral
and pelvic fins, the larvae developed pectoral fins,
although pelvic fins were absent (Fig. 2).
DISCUSSION
MORPHOLOGY
The general morphology of the pelvic fins and associated girdle in clariids closely resembles the
plesiomorphic condition in catfishes, with the central
plate of the basipterygium bearing two rostra1 processes (Shelden, 1937). Radial elements are absent.
This configuration can be found in both the fusiform
Clarias and the anguilliform Gymnallabes. In Channallabes the anterior processes appear to be absent
with the pelvic fin rays being supported by a simple
basipterygial element (Fig. 3a). In the C. apus specimens bearing both pelvic fins, both fins are supported
by such basipterygial elements. When both pelvic fins
are absent the basipterygial elements are also lacking
(Fig. 3c). The strict relation between fin development
02002 The Linnean Society of London, Biological Journal
and girdle development is demonstrated in a specimen
which lacks only the right pelvic fin: only the right
basipterygial element is absent. The left one has a
comparable morphology to that of normal girdle
(Fig. 3b).
The pectoral girdle in clariids is well developed, as
it is for most catfishes (Tilak, 1963). It consists of a
large cleithral bone, which is attached to a large coracoid bone. Both bones form ventral interdigitations
with their counterparts, thus fortifying the pectoral
girdle. Laterally the cleithrum bears a notch into
which fits the pectoral spine. This spine is generally
well developed in clariids, but may be small t o rudimentary in anguilliform species. Such a rudimentary
spine can be observed in Channallabes apus specimens that have pectoral fins (Fig. 4a-b). In those
specimens lacking a pectoral fin, no traces of a protruding pectoral spine can be observed (Fig. 4c-d). The
pectoral girdle, however, is still well developed with
strong midline suturing.
of
VARIABILITY IN LIMB REDUCTION AND
ANGUILLIFORMITY
The occurrence of variability in paired fin loss appears
to be related to the degree of anguilliformity. In the
DFA of all species root 1 represents the degree of
anguilliformity: lowest root-values correspond to the
fusiform body shape of Clarias gariepinus, whereas
highest root-values correspond to the extremely
elongated body of Dolichallabes rnicrophthalmus (Fig.
la). For all groups lacking pelvic a n d o r pectoral fins
the centroids have positive root 1-values (only Ga
has a negative value). Not only is the variability in
presencelabsence of paired fins observed only in
anguilliform species, this intraspecific variability is
the Linnean Society, 2002, 75, 367-377
372 D. ADRIAENS ETAL.
Figure 2. Larval Channallabesapus showing the well-developed pectoral fins (althoughthose of the parents were absent).
(A) Lateral view of the head, and (B) detail of the pectoral fin.
also related to the degree of anguilliformity: within all
three species showing this variance (i.e. Gymnallabes
alvarezi, Channallabes apus and D. microphthalmus),
the more paired fins that are absent, the higher the
root 1-values for the centroids of the subgroups.
From this DFA it also becomes clear that in the
(micro- and macro-) evolutionary process of body elongation, it is always the pelvic fins that are the first to
disappear. In the three species studied here which
show the intraspecific variation in limb development,
all have specimens lacking pelvic fins. All paired
fins are present in the less anguilliform specimens
of Channallabes apus (based on the position of the
centroid in the DFA). On the other hand, in the most
anguilliform specimens all paired fins are absent,
whereas in the intermediate group only the pelvic fins
are lacking; a situation that supports our hypothesis.
FINDEVELOPMENT AND FIN LOSS
In many teleosts lacking pelvic fins, the endoskeletal
pelvic girdle remains present (Nelson, 1994). Exceptionally, a large variation in the degree of pelvic fin
and girdle reduction has been observed independently
in many populations of several gasterosteid teleosts,
especially from lakes in recently deglaciated regions
(Nelson & Atton, 1971; Bell, 1987, 1994; Bell et al.,
1993). In these gasterosteids almost all degrees of
0 2002 The Linnean Society of London, Biological Journal
of
the Linnean Society, 2002, 75, 367-377
INTRASPECIFIC VARIATION IN LIMBLESSNESS
373
Figure 3. External (top row) and internal (two bottom rows) morphology of the pelvic fins and girdle in Channallabes
apus (ventral view). (a) Both pelvic fins present; (b) right pelvic fin present; and (c) pelvic fins both absent. Blue arrowhead indicates the position of the anterior margin of the anal fin; white arrowhead indicates the position of the anal
opening; and red arrowhead indicates the position of the pelvic girdle. Scale bar = 1mm.
morphological intermediates between fully developed
and absent structures were observed, although the
sequence in which elements of the pelvic complex
are lost is strikingly similar among populations (Bell,
1987). In Channallabes apus, however, this is not the
case. Specimens with two pelvic fins have their pelvic
fin rays supported by a basipterygial element. The loss
of a fin also implies the loss of the entire girdle-half
a t that side. Such a condition is not present in the
pectoral girdle, as it always retains its normal form
when the pectoral fin is absent. The fact that no intermediate stages were observed and that the pelvic
fins in mature C. apus are very small (compared to
other clariids), indicates that a minimal threshold of
relative size might exist. This has been observed in
tetrapods (Lande, 19781, and appears to be the case
for C. apus as well, a s the pelvic length did not go
below 13%of the skull length (nor did it for the other
clariid species studied). Such a threshold was not
observed in the pectoral fin (its size ranged between
1%and 50% of the skull length).
The difference in the degree of development in the
pelvic and pectoral fins and girdles may be the result
of a different timing of a triggering mechanism, which
inhibits fin formation in relation to functional demands. In teleosts it has been demonstrated that
the chondrogenic condensation of the pelvic girdle
immediately results in a n adult pattern, whereas the
pectoral girdle experiences a kind of larval metamorphosis (Grandel & Schulte-Merker, 1998). In relation
to functional demands and natural selection the complete loss of the pelvic girdle, in contrast to the conservation of the morphology of the pectoral girdle, is
to be expected. Although some secondary functions of
the pelvic fins such as in the formation of an egg pouch
are known in some catfishes, its primary function is
restricted to supporting the pelvic fin (Kohda et al.,
1995). The pectoral girdle, however, is of more importance in other crucial biological roles. The pectoral
girdle participates in the mechanics of buccal expansions associated with respiration and feeding, as well
as protecting the heart (Lauder & Liem, 1989; Liem,
1991). This multi-functional difference may be reflected in the fact that the first paired fin to disappear
is always the pelvic. In gasterosteids, the reduction
of pelvic elements (especially the spine) could be related to the absence of predatory fishes, but in a
complex way involving other factors like calcium levels
a s well (Bell et al., 1993). In clariids, predation will be
of little importance, since the pelvic fins have no protective function. A link with its life style may be more
plausible (see below).
0 2002 "he Linnean Society of London, Biological Journal of the Linnean Society, 2002, 75, 367-377
374
D. ADRIAENS ET AL.
Figure 4. Morphology of the pectoral fin in Churznullabes apus. (a, b ) Cleared and stained specimen, showing a small
pectoiA fin with spine and well-developed girdle. Black arrow indicates radials, white arrow indicates pectoral spine.
1 c. d 1 Exto-nal rnorphology of a specimen lacking pectoral fins. Scale bar = 1mm.
The hreeding results indicated t h a t t h e genetic
determination of pectoral fin loss may be variable,
since the larvae had well developed pectoral fins
although both parents did not. To what degree t h e
pectoral tin development as well as t h e minimal pelvic
fin length threshold is regulated genetically requires
further study. A partial genetic control of intraspecific
pelvic fin variability has, however, been demonstrated
in gasterosteid species (Nelson, 1977; Bell ef a!.,1993).
By analogy with snakes and lizards limb reduction
in anguilliform clariids may be attributed to body
elongation (Greer, 19901,which is associated with the
expansion of Hox-gene expression domains along the
body axis (Cohn & Tickle, 1999).If so, a significant dif-
C 2002 The Linnean Society of London, Biological Journal o f t h e Linricari Society, 2002. 75. 367-377
INTRASPECIFIC VARIATION IN LIMBLESSNESS
ference with snakes would still exist a s in the latter
body elongation involves thoracic segment multiplication, whereas in clariids it involves an increase in
caudal vertebrae (Cabuy et al., 1999). The result of the
breeding experiment may suggest one possible hypothesis, namely, that the pectoral fins regress during later
ontogeny. However, this seems very unlikely based on
the well-developed status of the fins in the larvae
(Fig. 2). Developmental regression has been observed
in some tetrapods but a t a much earlier stage: in
lizards and cetaceans the pelvic limb buds are initially
formed, after which they undergo a regression (Lande,
1978).
ADAPTATIONS TO A SPECIALIZED LIFE STYLE
What could be the driving force towards fin reduction
in clariids? The answer presumably can be found in
relation to their life style. A correlation between body
elongation (by means of an increasing number of
vertebrae (Hoffstetter & Gasc, 1969)) and limblessness has been demonstrated for several vertebrate lineages (Gans, 1975; Lande, 1978; Withers, 1981). It has
even been suggested that in lizards body elongation
preceded limb reduction during evolution, which
appears to be the case in Clariidae as well (Pellegrin,
1927; Gans, 1975). In general, limblessness and body
elongation have become associated with a burrowing
or fossorial behaviour (especially head-first burrowing) where additional structural adaptations may
involve the complete body musculature (Lee, 1998;
O’Reilly et al., 1997). In anguilliform clariids the axial
skeleton has an analogous structural advantage to
that of tetrapod Caecilia, i.e. the ribs do not enclose
the viscera latero-ventrally. In Clariidae body elongation involves the addition of caudal vertebrae, which
do not support ribs that would stress the viscera
during undulation (Cabuy et al., 1999). This is clearly
in favour of extreme anguilliform or undulatory locomotion, which facilitates both aquatic and terrestrial
locomotion (Gillis, 1996). Its beneficial result on
energy management has also been suggested (Gans,
1975, 1986). The efficiency of terrestrial locomotion in
Channallabes apus was observed in the field by one of
the authors (D.A). Once on the soil, specimens progressed very quickly through vegetation by means of
lateral undulations. However, it also appears that the
driving force towards body elongation and limblessness in this species is related to burrowing adaptations as well, This species is found in Northern Gabon,
to some degree living under similar conditions as
Protopterus, the African lungfish (both species are
syntopic (Teugels et al., 1991)). Channallabes apus
appears to live in the moist soil of the rain forest of
Central Africa where they could be caught by placing
bait in a small, hand-made furrow (specimens were
375
caught both during dry and rainy seasons). Within five
to 20 minutes after placing the bait (earth worms) on
a fish line, specimens were caught. This would indicate that C. apus is able to detect prey when it is subterranean and swimsfcrawls through the mud to catch
it. During the dry season, when the swampy regions
are largely dried up, water is restricted to below the
surface layer of the soil (where the water presumably
is in contact with small creeks in the neighbourhood).
During the rainy season the area is covered with tiny,
extremely shallow creeks and the soil is much more
saturated. This ecological niche differs from that
utilized by Protopterus, which is more restricted to
dryer regions (closer to savannas).
Other adaptations to a fossorial habit include
reduction of the eye, and may involve special modifications in the cranial morphology (for example skull
miniaturization) (Cabuy et al., 1999; Devaere et al.,
2001). Comparable body elongation as well as cranial
adaptations, combined with a similar burrowing behaviour, have been suggested for some Asian clariids
(Encheloclarias spp.) (Ng & Lim, 1993).
Variation in limb development in vertebrates is
generally considered to represent major structural
changes a t the macroevolutionary level where such
changes are believed to have a different basis than
those a t the microevolutionary level. In some species
of the African catfish family Clariidae, however,
variation in fin development appears to exist a t
the intraspecific level, thus representing a unique
example of microevolutionary variability. In these
species fin andlor girdle reduction can be linked to
body elongation, which is also the case in most other
elongated vertebrates. The process of paired fin loss in
clariids always involves the initial loss of the pelvic
fins, which may be followed by the loss of the pectoral
fins in Channallabes apus. The loss of the pelvic fins
is accompanied by the loss of the pelvic girdle. This is
not the case for the pectoral girdle. This may well be
a reflection of the differences in multi-functional
importance of the girdles. As is the case for most limbless and elongated vertebrates, the loss of fins in
clariids may possibly be related to the highly specialized fossorial life style.
ACKNOWLEDGEMENTS
The authors wish to thank the curators of fish collections a t the BMNH, MNHN, MHNG, NMW, MCZ,
ZMB and RMNH for the use of specimens under their
care. Special thanks for assistance in the statistic
evaluation of the data matrix t o A. Vanreusel, S.
Degraer and J. Vanaverbeke. We thank G. Howes
for reading and commenting the manuscript. Special
02002 The Linnean Society of London, Biological Journal o f t h e Linnean Society, 2002, 75, 367-377
376
L). ADRIAENS ET AL.
~
t h a n k s to the I F.A.F. (Gabon) for assistance during
sampling Research was funded by the BOF of the
Ghent liniversity (project 01104299) and the FWO
1 pTOJf?Ct c* 0388.00)
REFERENCES
Bell MA. 1987. Interacting evolutionary constraints in pelvic
reduction of threespine sticklebacks, Gasterosteus aculeatrts
(Pisces, Gasterosteidae). Biological Journal of the Linnean
.y(Jl'/Cf,V31 ( 41: 347-382.
Bell MA. 1994. Pelvic reduction in threespine stickleback from
('ook 1nlc.t lakes: geographical distribution and intrapopulat i o i i variation. ('ope7a 2: 314-325.
Bell MA, Orti G, Walker JA, Koenings JP. 1993. Evolution
of pelvic rcxduction in threespine stickleback fish: a test of
competing hypotheses. Ecolutiori 47 (3J: 906-914.
Boulenger GA. 1908. A revision of the African silurid fishes
of the subfamily Clariinae. Proceedings of the Zoological
Sf)(';L't,V (L(J/ltl(J//,2: 1062-1096.
Boulenger GA. 1911. Cutulogue of the fiesh-water fishes of
Afrrccr i n t h t ' British Museum (Natural History). London.
Printed by order of the Trustees. (2 edn).
Cabuy E, Adriaens D, Verraes W, Teugels GG. 1999. Comparative study on the cranial morphology of G.ymnal1ahe.s
t v p u s {Siluriformes: Clariidae) and their less anguilliform
re I a ti ves, ('Itr rirr iln h m n i d a s and Claria s gariepin us.
,Jo/irl/o/01 lbf(JrphOlOg? 240: 169-194.
Carroll RL. 1997. The fossil record of elongation and limb
reduction as a model of evolutionary patterns. Journal of
;%'forp/?o/lJg~232 ( 31: 241.
Carroll SB. 2001. The big picture. Nature 409 (68211: 669.
Cohn MJ, Tickle C. 1999. Developmental basis of limblessness a n d axial patterning in snakes. Nature 399 (67351:
4 i4--479.
Devaere S , Adriaens D, Verraes W, Teugels GG. 2001.
Cranial morphology of the anguilliform clariid Chanrzallabes
opus 1 Guiitlier. 1873) (Teleostei: Siluriformes): adaptations
ted t o a powerrul biting'? Journal of Zoology (London)
I2 1: 235-250.
Duellman WE, Trueb L. 1986. Biology of anzphibians. New
York: McGraw-Hill Co.
Futuyma DJ. 1998. Euolufionary bio1og.y. Sunderland,
Massachtisett.s: Sinauer Associates, Inc. (3rd edn).
Gans C. 1973. Locomotion and burrowing in limbless
vert ebra tcs. .
\'a t II rc 242 ( 5397 J: 4 14-4 15.
od limblessness: evolution and functional
11 Zoologi.<t 15: 455-467.
Gans C. 1986. Locomotion of limbless vertebrates: pattern and
c ~ V O ~ U t l o I 1Ht'lpc2tlJ/Ogicff
.
42 1): 33-46.
Gillis GB. 1996. Undulatory locomotion in elongate aquatic
vert c4~rates:anguilliform swimming since Sir James Gray.
.4??1Pr7f'U//ZfJdOgiSt 36: 656-665.
Grandel H, Schulte-Merker S. 1998. The development of
the pairtd [ins in the zebrafish (Danio rerio). Mechanisms of
/ ) ~ ~ l ~ c l i i / ~ /79:
r i ~ 99-120.
~~,t
Greer AE. 1987. Limb reduction in the lizard genus Lerista.
1. Variation in the number of phalanges and presacralc
vertebrae. Journal of' Herpetology 21 (41: 267-276.
Greer AE. 1990. Limb reduction in the lizard genus Leristri.
2. Variation in the bone complements of the front and rear
limbs and the number of postsacral vertebrae. Journal of
Herpetology 24: 142-150.
Hoffstetter R, Gasc JP. 1969. Vertebrae and ribs of mordern
reptiles. In: Bellairs, AA, Parsons, TS, eds. Biology of' reptiles. London: Academic Press, 201-310.
Kohda M, Tanimura M, Kikue-Nakamura M, Yamagishi
S. 1995. Sperm drinking by female catfishes: a novel
mode of insemination. Enuironmentol Biologv of Fis/~c,s42:
1-6.
Lande R. 1978. Evolutionary mechanisms of limb loss in
tetrapods. Evolution 32: 73-92.
Lauder GV, Liem KF. 1989. The role of historical factors
in the evolution of complex organismal functions. In: Wake.
DB, Roth, G, eds. Conzplex organisinnl functioiis: integmtioit
and evolution tn uertebratw. New York: Wiley & Sans,
63-78.
Lee MSY. 1998. Convergent evolution and character correlation in burrowing reptiles: towards a resolution of squamate
relationships. Biological Journal of the Linneon Socict,v
65 (4): 369-453.
Liem KF. 1991. A functional approach to the development
of the head of teleosts: implications on constructional
morphology and constraints. In: Schmidt-Kittler, N, Vogel.
K, eds. Constructional morphoiogx and ( ? L Y J / / I ~ ~ ~ Berlin:
J~.
Springer-Verlag, 23 1-249.
Motani R, You H, McGowan C. 1996. Eel-like swimming in
the earliest ichthyosaurs. Nature 382 (6589J: 347-848.
Nelson JS. 1977. Evidence of genetic basis for absence of'
the pelvic skeleton in brook stickleback, Culrtra inconstrrns,
and notes on the gcographical distribution and origin of the
loss. Journal of Fisheries Rpsearch Board of Crrnada 34:
1314-1320.
Nelson JS. 1989-90. Analysis of the multiple occurrence of
pelvic fin absence in extant fishes. Mats,ya 15-16: 21-38.
Nelson JS. 1994. Fishes of the Lcorld. 3rd edn. New York: dohn
Wiley & Sons, Inc.
Nelson JS, Atton FM. 1971. Geographic and morphological
variation in the presence and absence of the pelvic skeleton
in the brook stickleback, C u l a m inconsfans iKxtland 1, in
Alberta and Saskatchewan. Canadian Journal of Zoolog>
49: 343-352.
N g PKL, Lim KKP. 1993. The Southeast Asian catfish genus
Encheloclarias (Teleostei: ClariidaeJ. with descriptions of
four new species. fchthyological Explorafioras o f Freshuater
4 (11: 21-37.
O'Reilly JC, Ritter DA, Carrier DR. 1997. Hydrostatic
locomotion in a limbless tetrapod. Nafurc' 386 166221:
269-272.
Pellegrin J. 1927. La disparition des nageoiros pnires chez
les poissons Africains du groupe des clariines. Aniia(es Des
Sciences Naturelles, (Zoologic) Paris 10: 209-222.
Poll M. 1942. Description d'un genre nouveau de Clariidae
orignaire de Congo Belge. Rei,ue cl'e Zoologic c't de Botaniyue
Africaines 36 (1):96-100.
G 2002 The Linnean Society of London, Biological eJournal of the Linneaii Societv, 2002. 75
367-377
INTRASPECIFIC VARIATION IN LIMBLESSNESS
Pough FH, Andrews RM, Cadle JE, Crump ML, Savitzky
AH, Wells KD. 1998. Herpetology. New Jersey: Prentice
Hall, Inc.
Shelden FF. 1937. Osteology, myology, and probable evolution
of the nematognath pevic girdle. Annals of the New York
Academy of Sciences 37: 1-96.
Taylor WR, Van Dyke GC. 1985. Revised procedures
for staining and clearing small fishes and other
vertebrates for bone and cartilage study. Cybium 9 (2):
107-119.
Teugels GG. 1986.A systematic revision of the African species
of the genus Clarias (Pisces, Clariidae). Annales de Musee
377
Royal de L'afrique Centrale, Sciences Zoologiques 247:
1-199.
Teugels GG, Snoeks J, De Vos L, Diakanou-Matongo JC.
1991. Les poissons du bassin inferieur du Kouilou (Congo).
In: Dowsett FLJ, Dowsett-Lemaire F, eds. Flore et faune du
bassin du Kouilou (Congo) et leur exploitation. Tauraco Press
& CONOCO, 109-139.
Tilak R. 1963. Studies on the Nematognathine pectoral grdle
in relation t o taxonomy. Annals of the Museum of Natural
History 13 (6): 145-155.
Withers PC. 1981. Physiological correlates of limblessness
and fossorial lizards. Copeia 1: 197-204.
0 2002 The Linnean Society of London, Biological Journal of the Linnean Society, 2002, 75, 367-377