Content uploaded by James S Patterson
Author content
All content in this area was uploaded by James S Patterson on Aug 24, 2016
Content may be subject to copyright.
132 South African Journal of Science 101, March/April 2005 Research in Action
Preliminary study of wing
morphometry in relation to
tsetse population genetics
J.S. Patterson*and C.J. Schofield*†
COMPARATIVE MORPHOMETRIC ANALYSIS
of shape variation in the wings of different
tsetse species reveals close accordance
with the phylogenetics of these species indi-
cated by DNA sequence analysis. In practice,
the morphometric analysis is economical and
simple to carry out, suggesting that this could
become a useful surrogate or complementary
tool for large-scale studies of tsetse population
genetics, designed to identify discrete popu-
lation targets amenable to local elimination.
Introduction
Control of tsetse (Diptera, Glossinidae:
vectors of African trypanosomiasis) can
be achieved through a variety of tech-
niques, including traps, insecticide-
impregnated targets, live-baits, sequen-
tial aerial spraying (SAT), and sterile male
release (SIT) (see Hargrove1for review). In
most cases, however, the tsetse popula-
tions then tend to recover – either due to
flies surviving the initial interventions,
or immigration of flies from untreated
regions, or both. To achieve and sustain
local elimination of a target fly population,
it is therefore preferable to define the area
of intervention to include an entire
panmictic fly population, such that natural
immigration from neighbouring localities
is of low likelihood. This is most readily
achieved for isolated island populations,
as shown by the elimination of Glossina
pallidipes from the Island of Principe in
1914,2the eradication of G. pallidipes and
G. m. morsitans from Antelope Island,
LakeKariba,Zimbabwe, in 1984,1,3 and the
elimination of G. austeni from Unguja
Island of Zanzibar in 1997.4But for most
mainland populations of tsetse, the geo-
graphical limits of target tsetse popula-
tions are less easily definable.
Application of population genetics
techniques can reveal the existing level of
population differentiation in tsetse,
providing guidance on the distribution of
genetically defined sub-populations. In
essence, the population genetics models
are used to estimate rates of gene flow
between populations, which are taken as
a surrogate for the rate of migration of
individuals. Allozyme studies, for example,
have revealed high levels of genetic
differentiation within populations of
G. pallidipes and other species of the
morsitans group in East Africa,5suggest-
ing that these species exist as a series of
relatively isolated populations, each of
which might be targeted separately for
control interventions. Similarly, mito-
chondrial and microsatellite DNA analy-
ses also reveal a high level of population
structuring within species of the morsitans
group in southern and eastern Africa,6,7
and within some of the G. palpalis gam-
biense populations in West Africa.8,9
Extensive further studies of population
structuring in tsetse seem appropriateas a
guide to planning progressive control
interventions, as envisaged by the
AU-PATTEC initiative.10 In addition, such
studies could help in post-control moni-
toring for analysing the likely source of
survivors or immigrants into treatedareas
– as shown for Triatominae, vectors of
American trypanosomiasis.11 To minimize
the use of expensive techniques of DNA-
sequence analysis for such studies, we
present here a preliminary comparison
of geometric wing morphometry as an
inexpensive surrogate for genetic analy-
sis. This work was carried out at the level
of tsetse species and species-groups, as a
prelude to further studies of within-
species differentiation.
Materials and methods
The insects. Samples were received as
individuals or groups of flies in 70%
ethanol, from colonies maintained at the
FAO/IAEA laboratories in Seibersdorf,
Austria, and from the CIRDES laborato-
ries at Bobo Dioulasso, Burkina Faso.
Additional samples of G. p. gambiense
were collected by trapping along the Kou
valley, Burkina Faso (Table 1).
Wing morphometry. Wings were removed
and dry-mounted between two micro-
scope slides. The right wing of each speci-
men was photographed using a digital
camera. Images of each wing were subse-
quently digitized and 7 cartesian coordi-
nates (homologous landmarks defined by
vein intersections; Fig. 1) were recorded
automatically using TPSdig software
(version 1.39).12 The x, y coordinates were
subjected to generalized procrustes anal-
ysis (GPA)13 and subsequently to a
thin-plate spline analysis14 using TPSrelw
software (version 1.35)15 and TPS regr
(version 1.26),16 allowing visualization of
shape differences as deformation grids.
The analysis produces variables subdi-
vided into uniform and non-uniform
components of shape changes. To offset
the problem of small sample sizes, a prin-
cipal component analysis of the shape
variables delivers fewer shape compo-
nents (‘relative warps’), explaining most
of the shape variance within the data set.
The relative warps were subsequently
analysed by discriminant analysis. Size
differences were assessed using centroid
size (CS), an isometric estimator of size
derived from the GPA superimposition
procedure. Finally, mean Mahalanobis
distances were used in a cluster analysis to
construct a UPGMA dendrogram (Un-
weighted Pair Group Method with Arith-
metic Mean). Multivariate analyses and
graphs were completed using JMP®ver-
sion 4.0.5 (SAS Institute Inc. 2001) and
Intercooled STATA 8.2 for Windows (Stata
Corporation 2003)
DNA sequence comparison: For comparison
with the wing morphometry, we used
available partial sequences of the ribo-
somal DNA internal transcribed spacer-2
(ITS2) downloaded from GenBank
(http://www.ncbi.nlm.nih.gov/) (Table 1).
These sequences were aligned using
Clustal-X and analysed by Neighbor-
*Department of Infectious and Tropical Disease, London
School of Hygiene and Tropical Medicine, London WC1
E7HT, U.K.
†Author for correspondence.
E-mail: cj.schofield@lshtm.ac.uk Fig. 1. Slide-mounted tsetse wing showing landmarks used for morphometric analysis.
Research in Action South African Journal of Science 101, March/April 2005 133
Joining using the Kimura-2-parameter
model of base substitution to construct
phylogenetic trees with 1000 bootstrap
replicates.
Results
The first five relative warps (shape
components) accounted for 90.6% of the
variance in the total data set, and were
used as input for the discriminant analysis.
Regressing centroid size against the first
relative warp gave no significant correla-
tion (r= 0.002), suggesting that size is not
the primary factor influencing the major
shape differences. Figure 2 shows the
discrimination of the three main species
groups by wing morphometry. The
discriminant model gave correct reclassi-
fication scores of 100% for pooled mem-
bers of the fusca and morsitans groups
and 95.2% for palpalis (4.2% assigned to
morsitans group). These reclassification
scores were ‘almost perfect’ (kappa =
0.96).17 The first two canonical vectors
(CV) together accounted for 84% of the
total heterogeneity (CV1–56% and
CV2–28%). The thin-plate spline repre-
sentations (Fig. 2a,b) show that most of
the shape change is associated with a
relative elongation of the wing, and this
separates fusca from the morsitans and
palpalis species groups. Figure 2c,d show
that the secondary factor of shape change
is related to the relative arrangement of
vein junctions, and clearly discriminates
between the morsitans and palpalis
groups.
Analysis of the ITS2 sequences revealed
three major clades, with good bootstrap
support. These correspond to the three
species groups, and show clear congruence
with a cluster analysis of the morpho-
metric data (Fig. 3).
Discussion
The 31 currently recognized species and
subspecies of Glossina are customarily
placed into three species groups which
are sometimes given subgeneric status18 –
the fusca group (subgenus Austenina),
palpalisgroup (subgenus Nemorhina), and
morsitans group (subgenus Glossina).
These groupings are based primarily on
morphological features of the adult
genitalia,19 although they also reflect
differences in distribution, habitat and
behaviour.20 Species of the fusca group
typically occur in lowland rain forests of
Westand Central Africa (exceptions being
G. longipennis and G. brevipalpis in drier
regions of eastern Africa); species of the
palpalis group are more usually associ-
ated with riverine vegetation, but also
extend into savanna regions between
river systems; while species of the morsi-
tans group are primarily associated with
drier savannas. In this study, the geomet-
ric analysis of wing morphometry suc-
cessfully recovered not only the species,
but also the three species-groups.
As initially shown by Solano et al.8for
G. p. gambiense, a degree of correlation can
Table 1.
Glossina
specimens used in the study.
Species group Species Source of material for morphometry (n) ITS2 GenBank accession nos
morsitans G. morsitans s.l. FAO/IAEA, CIRDES (21) AF021360; AF021359; F021358
G. pallidipes FAO/IAEA (5) AF021357(1); AF021356(2)
G. swynnertoni FAO/IAEA (4) AF021355
palpalis G. palpalis gambiensis FAO/IAEA, CIRDES, and Kou valley (16) AF024505
G. fuscipes FAO/IAEA (4) AF021352
G. tachinoides CIRDES (4) AF021353
fusca G. brevipalpis FAO/IAEA (3) AF022361(1); AF022360(2)
Fig. 2. Discriminant analysis of the morphometric data, showing the distribution of specimens in the space defined by the first two canonical variates (CV1 and CV2).
a–d are thin-plate splines showing, by deformations from the mean, shape differences of the wings that correspond to the indicated species/species groups on both axes
(landmarks on the deformation grids are numbered as in Fig. 1).
134 South African Journal of Science 101, March/April 2005 Research in Action
be found between estimates of popula-
tion structuring based on analysis of
microsatellite DNA, and comparative
wing morphometrics. Using linear wing
morphometrics, these authors showed
clear separation between G. p. gambiense
populations of Senegal and Burkina Faso,
but not between populations within
Burkina Faso which were revealed by
comparative analysis of microsatellite
DNA sequences. Genetic separation of
the Senegal populations from those of
Maliwas also confirmed by Marquez et al.9
using comparisons of mitochondrial
DNA sequences. In our study, the very
high congruence over seven species and
subspecies between genetic comparisons
based on a ribosomal DNA sequence, and
phenetic comparisons based on geometric
morphometry, suggests that the geometric
analysis is a more sensitive surrogate
for the DNA sequence comparisons. It
appears, moreover, that wing shape may
represent a relatively neutral trait that is
not heavily modulated by ecological
adaptationor environmental constraints.
In practice, data collection for geometric
analysis of wing shape is relatively
simple. The wings can be dry mounted
between microscope slides and then
either scanned (using a computer scan-
ner) or photographed with a digital cam-
era. The resulting image can then be
processed using freely available software
(e.g. http://life.bio.sunysb.edu/morph/),
or sent as an e-mail attachment to a refer-
ence laboratory for further analysis. Such
a procedure is simpler and much less
costly than current techniques for DNA ex-
traction and sequencing, and offers op-
portunities for rapid processing of large
samples of field-collected material cover-
ing the entire distributional range of tar-
get species and subspecies of tsetse. This
would permit detailed analysis of popula-
tion structuring to identify the geograph-
ical limits of discrete or panmictic
populations, which would represent
targets for control interventions that
would be least likely to suffer from
post-control reinvasion. In addition, as
shown for Triatominae,11 such studies
can also be used for post-intervention
monitoring, providing a way to confirm
whether any newly encountered tsetse
are survivors from the initial control
interventions (indicating a local control
failure) or are immigrants from a neigh-
bouring population of that species (per-
haps indicating a breakdown of control
barriers).
In the context of the African Union
initiative to eliminate the problem of
tsetse and trypanosomiasis (AU-PATTEC),10
we believe that these techniques could
be particularly applicable for defining
geographical areas amenable to large-
scale elimination of the tsetse popula-
tions. There is a wealth of evidence that
tsetse control is feasible, but also that it is
difficult to sustain over the long term.1,21,22
By contrast, local elimination of tsetse is
sustainable, but generally held to be feasi-
ble only for geographically constrained
situations such as islands.1,2,4 The task of
population genetics studies is, in a sense,
to find and define those biogeographical
‘islands’ of tsetse distribution on main-
land Africa, and our study shows that this
may be feasible on a large scale using the
techniques of geometric morphometry,
with confirmation from smaller-scale
studies using the more expensive methods
of DNA sequence comparisons.
A summary of this work was presented as a poster at
the IX European Multicolloquium of Parasitology,
Valencia, Spain (18–23 July 2004). We thank Alan
Robinson (FAO/IAEA Laboratories), Abdoulaye
Gouro and Idrissa Kabore (CIRDES), and Tamboura
Issa (UCLT, Burkina Faso) for supplying the insects
used in this study, and Brian Williams for encourage-
ment and critical reading of the manuscript.
1. Hargrove J.W. (2003). Tsetse eradication: sufficiency,
necessity and desirability. Report to DFID, Edin-
burgh.
2. Da Costa B.F.P., Sant’anna J.F., Santos A.C. and
Alvares M.G.A. (1916). Sleeping Sickness, a record of
four years war against it in Principe, Portuguese West
Africa. Ballière, Tindall and Cox, London.
3. Hargrove J.W. and Williams B.G. (1998). Opti-
mised simulation as an aid to modelling, with an
application to the study of tsetse flies, Glossina
morsitans morsitans Westwood (Diptera: Glossini-
dae). Bull. Ent. Res. 88, 425–435.
4. Vreysen M.J.B., Saleh K.M., Ali M.Y., Abdullah
M.A., Zhu Z.R., Juma K.G., Dyck V.A., Masangi
A.R., Mkonyi P.M. and Feldmann H.U. (2000).
Glossina austeni (Diptera: Glossinidae) eradicated
on the Island of Unguja, Zanzibar, using the
sterile insect technique. J. Econ. Entomol. 93,
123–135.
5. Krafsur E.S. and Griffiths N. (1997). Genetic
variation at structural loci in the Glossina morsitans
species group. Biochem. Genet. 35, 1–11.
6. Wohlford D.L., Krafsur E.S., Griffiths N.T.,
Marquez J.G. and Baker M.D. (1999). Genetic
differentiation of some Glossina morsitans popula-
tions. Med. Vet. Entomol. 13, 377–385.
7. Krafsur E.S. (2003). Tsetse fly population genetics:
an indirect approach to dispersal. Trends Parasitol.
19, 162–166.
8. Solano P., De La Rocque S., Cuisance D., Geoffroy
B., De Meeus T., Cuny G. and Duvallet G. (1999).
Intraspecific variability in natural populations of
Glossina palpalis gambiense from West Africa,
revealed by genetic and morphometric analysis.
Med. Vet. Entomol. 13, 401–407
9. Marquez J.G., Vreysen M.J.B., Robinson A.R.,
Bado S. and Krafsur E.S. (2004). Mitochondrial
diversity analysis of Glossina palpalis gambiensis
from Mali and Senegal. Med. Vet. Entomol. 18,
288–295.
10. Kabayo J.P. (2002). Aiming to eliminate tsetse from
Africa. Trends Parasitol. 18, 473–475.
11. Dujardin J.P., Schofield C.J. and Panzera F. (2000).
Les Vecteurs de la Maladie de Chagas. Recherches
Taxonomiques, Biologiques et Génétiques.
Academie Royale des Sciences d’Outre Mer,
Brussels.
12. Rohlf F.J. (2003). TPS dig version 1.39. Department
of Ecology and Evolution, State University of
New York, Stony Brook, NY. Available from
http://life.bio.sunysb.edu/morph/.
13. Rohlf F.J. (1990). Rotational fit (procrustes)
methods. In Proc. Michigan Morphometrics Work-
shop, eds F.J. Rohlf and F.L. Bookstein, pp. 227–236.
University of Michigan Museums, Ann Arbor.
14. Bookstein F.L. (1991). Morphometric Tools for
Landmark Data: Geometry and biology, p. 435.
Cambridge University Press, Cambridge.
15. Rohlf F.J. (2003). TPS relw version 1.35. Department
of Ecology and Evolution, State University of
New York, Stony Brook, NY. Available from
http://life.bio.sunysb.edu/morph/.
16. Rohlf F.J. (2000). TPS regr version 1.26. Department
of Ecology and Evolution, State University of
New York, Stony Brook, NY. Available from
http://life.bio.sunysb.edu/morph/
17. Landis J.R. and Koch G.G. (1977). The measure-
ment of observer agreement for categorical data.
Biometrics 33, 159–174.
18. Haeselbarth E., Segerman J. and Zumpt E. (1966).
The arthropod parasites of vertebrates in Africa
south of the Sahara (Ethiopian region). 3 (Insecta
excl. Phthiraptera). Publications of the South African
Institute for Medical Research 13, 1–283.
19. Newstead R., Evans A.M. and Potts W.H. (1924).
Guide to the Study of Tsetse-flies. Hodder &
Stoughton, London.
20. Jordan A.M. (1993). Tsetse-flies (Glossinidae). In
Medical Insects and Arachnids, eds R.P. Lane and
R.W. Crosskey, pp. 333–388. Chapman & Hall,
London.
21. AHP (2002). Tsetse control: the next 100 years. DFID
Animal Health Programme, Centre for Tropical
Veterinary Medicine, Edinburgh.
22. AHP (2003). Recent advances in livestock keeper-based
tsetse control: the way forward. DFID Animal Health
Programme, Centre for Tropical Veterinary
Medicine, Edinburgh.
Fig. 3. Comparison of Neighbor-Joining k2p linearized tree (left) with UPGMA cluster analysis of Mahalanobis
distances from wing morphometrics (right). Numbers indicate per cent bootstrap support from 1000 replicate
analyses.
