Content uploaded by James Stuart F Barker
Author content
All content in this area was uploaded by James Stuart F Barker on Mar 30, 2018
Content may be subject to copyright.
Genetic identification of wild Asian water buffalo in Nepal
INTRODUCTION
There are at least 130 million domestic water buffalo
(Bubalus bubalis), and more human beings depend on
them than on any other domestic animal (Scherf, 2000).
Two types are recognized, based on morphological and
behavioural criteria (Macgregor, 1939) – the river buffalo
of the Indian subcontinent and further west to the Balkans
and Italy, and the swamp buffalo, found from Assam in
the west through southeast Asia to the Yangtze valley of
China in the east. Morphologically they are quite distinct,
with the swamp buffalo exhibiting white chevron, socks
and tip of tail, and relatively straight pale-coloured horns,
and river buffalo having a black body and generally
curved horns. They also differ in chromosome number
(Ulbrich & Fischer, 1967; Fischer & Ulbrich, 1968) –
swamp, 2n = 48, river, 2n = 50, and in allele frequencies
at protein coding (Amano, 1983; Barker et al., 1997b) and
microsatellite loci (Barker et al., 1997a). One exception
to the morphological distinction of the river and swamp
types is the native buffalo of Sri Lanka. Although
phenotypically resembling the swamp type, they have a
chromosome number of 2n = 50 (same as river-type), and
phylogenetic trees derived from protein coding and
microsatellite loci cluster them strongly with river-type
populations (Barker et al., 1997a,b). Morphologically, the
swamp type is more similar than the river type to the wild
Asian buffalo (Bubalus arnee (Kerr) 1792). Barker et al.,
(1997b) therefore suggested that the Sri Lankan buffalo
represents the ancestral type from which the present
breeds of river buffalo of the Indian subcontinent were
derived.
The genus Bubalus was widely distributed in Europe and
southern Asia in the Pleistocene, but later was restricted to
the Indian subcontinent and southeast Asia (Mason, 1974).
The time of divergence of the swamp and river types has
been estimated in a number of studies: using restriction site
variation in rDNA and mtDNA, Amano et al. (1994)
estimated the time of divergence as within the last 1.0 Myr
and 1.0–2.0 Myr respectively. Tanaka et al. (1995)
estimated at least 0.7 Myr, based on mtDNA restriction site
variation, while Tanaka et al. (1996), using cytochrome b
sequences, estimated 1.7 Myr. From data on 21
microsatellite loci in eight swamp and three river
populations, Barker et al. (1997a) estimated a minimum
divergence time of 10,000–15,000 years ago. An estimate
based on sequence data for the D-loop region of mtDNA
(Lau et al., 1998) was 28,000–87,000 years ago. Most of
these estimates are derived from mtDNA variation, which
Nei (2002) argues is not very suitable for estimating
divergence times. Further, the wide variation is probably a
J. R. B. Flamand1,5, D. Vankan2, K.P. Gairhe3, H. Duong2and J. S. F. Barker4
1Royal Chitwan National Park, Department of National Parks and Wildlife Conservation, PO Box 860, Kathmandu, Nepal
2Cattle DNA Typing Laboratory, Department of Veterinary Pathology, University of Queensland, St Lucia, Queensland 4072, Australia
3Department of National Parks and Wildlife Conservation, PO Box 860, Kathmandu, Nepal
4School of Rural Science and Agriculture, University of New England, Armidale NSW 2351, Australia
5Present address: PO Box 456, Mtubatuba 3935, South Africa
(Received 5 June 2002; resubmitted 17 December 2002; accepted 3 February 2003)
Abstract
The wild water buffalo is highly endangered, with the few remaining populations already affected or
likely to be increasingly affected by hybridization with domestic buffalo. The work described here was
done to evaluate a genetic method to discriminate wild from mixed ancestry (hybrid) and domestic
animals, and to identify with high probability those most likely to be purebred wild. Samples from 45
animals (phenotypically classified into three groups – ten wild, 28 domestic and seven hybrid) were
genotyped for ten microsatellite loci. Although genetic distances among the three groups were small, an
assignment test identified two of the ‘wild’ and seven of the ‘domestic’ as hybrids. However, sample
sizes also are small, indicating the need for a conservative approach in the first instance in using these
results. As more animals are genotyped, assignments will become more accurate, and a translocation
programme to establish a second Nepalese wild population in a protected area could be undertaken.
All correspondence to: J. S. F. Barker. Tel.: ++ 61 2 6773 3924;
Fax: ++ 61 2 6773 3275; E-mail: sbarker@metz.une.edu.au.
function of the different rates of evolution of the genes and
sequences used, while the longer estimates which are based
on single genes are probably overestimates as the time of
gene splitting may be much earlier than the time of
population splitting (Takahata & Nei, 1985). Nevertheless,
divergence of the two types clearly preceded their
domestication, as water buffalo apparently were
domesticated in India some 5000 years ago, and in China
4000 years ago (Cockrill, 1974), although Chen & Li
(1989) suggested domestication in China at least 7000 years
ago. Lau et al. (1998) hypothesized that the water buffalo
originated in mainland southeast Asia, and then spread
north to China and west to the Indian subcontinent where
animals of the river type evolved. Much later, and from
each of the lineages in China and the Indian subcontinent,
the swamp and river types were domesticated.
The wild buffalo is highly endangered, with a world
population considered by FAO to be certainly fewer than
4000 animals, possibly fewer than 200, and it is even
possible that no purebred wild animals exist (Scherf, 2000).
Scherf (2000) indicated that small isolated populations may
remain in the Kosi Tappu Wildlife Reserve (Nepal), Bastar
and Raipur Districts of Madhya Pradesh and Manas
Wildlife Sanctuary/ Project Tiger Reserve (India), Royal
Manas National Park (Bhutan), and Huai Kha Khaeng
Wildlife Sanctuary (Thailand), with these populations
believed to have been least affected by interbreeding with
domestic and/or feral buffalo. Heinen & Singh (2001)
censused wild buffalo in Kosi Tappu Wildlife Reserve, and
estimated a population of 145 wild buffalo, and a highly
backcrossed, semi-feral population of 131 animals. These
authors conclude: ‘Given the global and national status of
this species as documented by IUCN’s Asian Wild Cattle
and Buffalo Specialist Group and the American Zoo
Association, a translocation to begin a second Nepalese
population is of the highest conservation priority.’
A major component of the threat to the wild buffalo is
continuing hybridization with domestic buffalo. If a
second Nepalese population is to be established in a
protected area, every effort should be made to ensure that
this foundation population comprises only purebred wild
animals. The domestic buffalo in the vicinity of Kosi
Tappu are river type, and thus phenotypically distinct from
wild animals. But animals of mixed wild–domestic
ancestry may not be phenotypically distinguishable from
wild animals.
However, as swamp and river buffalo are significantly
differentiated genetically (Barker et al., 1997a,b), the wild
buffalo and domestic river buffalo in Nepal could be
sufficiently genetically differentiated that highly
polymorphic molecular markers would allow the assignment
of individual animals to wild, domestic or ‘hybrid’ classes.
Here we test this expectation using ten microsatellite loci on
a sample of animals from Kosi Tappu, Nepal.
MATERIALS AND METHODS
Hair samples were collected from the tails of 45 water
buffalo in or in the vicinity of Kosi Tappu Wildlife
Reserve, Nepal, and sent to the Cattle DNA Typing
Laboratory, University of Queensland, for microsatellite
assay. Three groups were defined, with the animals
sampled classified as wild (n= 10), domestic (n= 28) and
mixed wild/domestic ancestry (referred to here as hybrid:
n= 7). This classification was based on location of
sampling (in the wild or in villages) and behavioural and
phenotypic traits. All animals classified as domestic were
river type with black body and curled horns (as in the
Murrah breed river buffalo), while those classified as wild
had white chevron, socks and tip of tail, and larger,
relatively straight, pale-coloured horns (similar to swamp
buffalo) (Heinen, 2002). Two animals (98/01 and 99/01)
were identified by villagers as F1hybrids, with 98/01 the
more certain as an offspring of a known domestic cow
(97/01) and a wild bull.
Ten microsatellite loci were assayed (all prefixed
CSSM – 19, 29, 32, 33, 38, 41, 43, 57, 60 and 61), all of
which have been used previously to characterize
populations of water buffalo in southeast Asia (Barker et
al., 1997a). Methods used for DNA extraction, PCR and
genotype scoring were as given by Barker et al. (1997a).
Analyses of genetic variation
Allele frequencies, tests for Hardy–Weinberg equilibrium
and genotypic differentiation among the three groups
(wild, hybrid and domestic) were estimated using the
computer program GENEPOP 3.3 (Raymond & Rousset,
1995). Significance levels for the Hardy–Weinberg tests
were determined by applying to the probability estimates
calculated by GENEPOP the sequential Bonferroni
procedure (Hochberg, 1988; Lessios, 1992) over loci
within each group. Expected heterozygosities at each
locus in each group, and F-statistics were estimated using
FSTAT 2.9.3 (Goudet, 1995), with the sequential
Bonferroni procedure applied over loci in deriving
significance levels. These parameters of population
structure, computed using the methods of Weir &
Cockerham (1984), are defined as the correlations
between pairs of genes (1) within individuals (F), (2) of
different individuals in the same population (
θ
), and (3)
within individuals within populations (f), and are
analogous to Wright’s (1951, 1978) FIT, FST and FIS
respectively. Genetic distances among the groups
(standard genetic distance of Nei, 1978) were obtained
using the DISPAN computer program (T. Ota, pers.
comm.). Genetic distances among individuals were
estimated as 1 – proportion of shared alleles (DPS,
Bowcock et al., 1994) with MICROSAT 1.5d (Minch et
al., 1995). This distance matrix was then used to construct
a neighbour-joining tree (Saitou & Nei, 1987), using the
program NEIGHBOUR in PHYLIP 3.5c, available at
http://evolution.genetics.washington.edu/phylip.html.
Genetic admixture analysis and population
assignment
The primary objective was to discriminate wild from hybrid
and domestic animals, i.e. to identify with high probability
those animals most likely to be purebred wild. The
assignment method of Pritchard, Stephens & Donnelly
(2000) was used, as we wish not only to determine whether
an individual animal is wild, hybrid or domestic, but also
to estimate individual mixture proportions. Where sampled
individuals are subdivided into predefined groups, it is
possible to test for each individual whether it should be
assigned to the group to which it was predefined, or whether
it is the result of immigration in the present or in a previous
generation from another group. This method (implemented
in the program STRUCTURE – http://www.stats.ox.ac.uk/
~pritch/home.html) is a Bayesian clustering approach using
multi-locus genotypes to infer population structure and
simultaneously to assign individuals to populations, and
assumes Hardy–Weinberg equilibrium and linkage
equilibrium between loci within each population. It assumes
there are Kpopulations (where Kmay be unknown), each
of which is characterized by the allele frequencies at each
locus. Individuals are assigned probabilistically to
populations, or jointly to two or more populations if their
genotypes indicate that they are admixed. For this analysis,
a value must be specified for the parameter vof Pritchard
et al. (2000) – the probability that an individual is
misclassified or has an ancestor(s) from other populations.
Several independent runs were done at different values of
vto determine if conclusions were robust to choice of v. As
the three predefined groups were only weakly genetically
differentiated (see Results), the model used assumed allele
frequencies in the groups were correlated. A burn-in
period of 200,000 steps was used, followed by 2,000,000
MCMC replicates.
RESULTS
Analysis of genetic variation
All microsatellites were polymorphic in all three groups,
with 3–11 alleles per locus (average 6.5 ± 2.4), and
expected heterozygosity (He) estimates ranged from 0.472
to 0.822 in wild, 0.548 to 0.821 in hybrids, and 0.417 to
0.858 in domestic (Table 1). Mean Hewas highest for the
hybrids, which had the lowest mean number of alleles, but
differences among the groups were not significant.
With these small sample sizes, alleles at low frequency
could by chance be found in one group but not the others.
At a threshold frequency of 5%, there were two, nought
and 13 private alleles in wild, hybrid and domestic buffalo
respectively (for a 10% threshold, numbers were three,
one and 15 respectively).
No loci showed significant deviations from
Hardy–Weinberg equilibrium in the wild or hybrid
groups. For the domestic, three loci (CSSM29, CSSM32
and CSSM57) showed an apparent significance of excess
homozygotes, but none was significant after Bonferroni
correction. None of the F-statistics for individual loci was
significant, but the means over loci were significant for all
three statistics (Table 2). Pairwise
θ
estimates were
significant only for the wild/domestic comparison (0.028,
P< 0.05). Tests of multi-locus population genotypic
differentiation (GENEPOP) showed wild and domestic to
be significantly different (P< 0.001), and also hybrid and
domestic (P< 0.05).
The neighbour-joining tree of individuals showed three
major clusters, but these did not correspond closely to the
three predefined groups. Two of these clusters each
included a subcluster of wild and hybrid individuals, and
the third cluster comprised primarily domestic animals.
As expected from this result, Nei standard genetic
distances among the predefined groups were small and not
significantly different from zero: wild/hybrid = –0.0002
± 0.0204, wild/domestic = 0.0504 ± 0.0284 and
hybrid/domestic = 0.0447 ± 0.0385.
Linkage equilibrium
No useful tests for linkage equilibria could be obtained
from this small data set. However, analyses were done on
the data of Barker et al. (1997a), comprising eight swamp
and three river buffalo populations. The only significant
disequilibria were for CSSM19/CSSM32 and for
CSSM32/CSSM43 (both P< 0.001). No information is
available on the chromosomal locations of these loci in
water buffalo, but the cattle–buffalo chromosome
homologies have been described (Di Berardino &
Table 1. Microsatellite loci used and summary of allelic variation in
the preclassified animals – ten ‘wild’, 28 ‘domestic’ and seven ‘hybrid’
buffalo
Locus No. Size Wild Hybrid Domestic
alleles range
(total)
No. HeNo. HeNo. He
alleles alleles alleles
CSSM19 6.7 131–151 3.7 0.472 3.7 0.655 5.7 0.626
CSSM29 3.7 184–188 3.7 0.478 3.7 0.583 3.7 0.555
CSSM32 5.7 208–224 3.7 0.472 3.7 0.631 5.7 0.542
CSSM33 8.7 154–173 5.7 0.622 5.7 0.810 7.7 0.642
CSSM38 9.7 163–189 6.7 0.756 3.7 0.571 6.7 0.533
CSSM41 5.7 122–144 3.7 0.472 3.7 0.571 5.7 0.417
CSSM43 5.7 222–254 3.7 0.528 3.7 0.560 5.7 0.744
CSSM57 5.7 120–128 4.7 0.589 2.7 0.548 5.7 0.623
CSSM60 8.7 95–135 5.7 0.650 5.7 0.738 7.7 0.711
CSSM61 11.7 92–128 8.7 0.822 7.7 0.821 11.7 0.858
Mean .76.5 4.3 0.586 3.7 0.649 5.9 0.625
SE .7 2.4 1.7 0.126 1.5 0.105 2.1 0.125
Heexpected heterozygosity
Table 2. F-statistics analysis for ten microsatellite loci in three
preclassified populations of water buffalo, with significance estimates
from permutation tests in the FSTAT programme
Locus f(FIS)θ(FST)F(FIT)
CSSM19 –0.005 0.029 0.024
CSSM29 0.305 –0.024 0.288
CSSM32 0.219 –0.020 0.203
CSSM33 –0.072 0.041 –0.027
CSSM38 0.090 0.028 0.115
CSSM41 0.163 –0.025 0.142
CSSM43 0.003 0.053 0.056
CSSM57 0.157 –0.032 0.130
CSSM60 0.017 0.034 0.050
CSSM61 0.051 0.072 0.120
Meana0.082 (0.035)** 0.022 (0.012)** 0.102 (0.027)**
aStandard deviations in parentheses – estimate from jackknife over loci
** P< 0.01
Iannuzzi, 1984). In cattle, CSSM19 and CSSM32 are
located on chromosome 1 (20 cM apart), and CSSM43
on chromosome 27 (Barendse et al., 1994; Moore et
al., 1994).
Genetic admixture analysis and population
assignment
For the STRUCTURE analysis, assignment of individual
animals was done using prior population information
(option USEPOPINFO = 1). That is, we assume that each
individual belongs with high probability to the group to
which it was classified (wild, hybrid or domestic), but
allow some probability that it has ancestry in the other
groups. Intergroup ‘migration’ rate (vin Pritchard et al.,
2000) was set at 0.05 or 0.1; that is, at the higher end of
the range recommended by Pritchard et al. (2000) in order
to be conservative in assignment. Results were slightly
sensitive to the value of v, with somewhat lower
probabilities of assignment to the predefined group at the
higher value of v. We report results for v= 0.05 (Table 3),
but some caution is necessary in interpreting them. That
is, there may have been slightly more mixing and
consequently individuals accepted here as wild that have
some ancestry from hybrid or domestic animals.
Estimated ancestry values in Table 3 are the posterior
values of qi(i= 1,2,3) for each individual, or the
proportion of each individual genotype that derives from
one or more clusters. We have taken a qvalue of 0.9 as a
conservative cut-off point (or significant value for q> 0.9).
That is, animals initially classified on phenotypic criteria
to the wild or domestic clusters, whose qvalue for that
cluster was ≥0.9, were accepted in that cluster, while if
q< 0.9, the animal was assigned to the hybrid cluster.
Thus eight of the ten predefined wild animals were
assigned to the wild cluster with q1≥0.93 (mean 0.971).
Two animals (27 and 69/01) that were predefined as wild
were assigned to the wild cluster, but showed substantial
ancestry from domestic (0.117) and hybrid (0.132)
respectively. All of the predefined hybrids were assigned
to the hybrid cluster, the lower q2values for some animals
(32, 35 and 73/01) indicating higher proportions of either
wild or domestic ancestry. Twenty-one of 28 predefined
domestic animals were assigned to the domestic cluster
with q3≥0.91 (mean = 0.965). Seven animals that were
predefined as domestic show substantial ancestry from the
wild or hybrid groups (see Table 3).
Thus we take these seven animals and the two wild
(27 and 69/01) as misclassified, and redefine them as
hybrid. This data set was then run with these three new
groups defined (USEPOPINFO = 1), confirming the
classification, that is, all qi> 0.9 (Table 3(b)).
As the structure method assumes linkage equilibrium,
the analysis was rerun after deleting locus CSSM32 from
the data. Only one change in assignment resulted –
one domestic animal whose q3was less than the 0.9
cutoff (0.893) for all ten loci now had q3= 0.930. Given
this minor difference, analyses based on all ten loci have
been retained.
Finally, a STRUCTURE analysis was done with prior
information on the wild and domestic animals only, and
not on the hybrids. That is, the data on the wild and
domestic animals are used to infer the ancestry of the
presumed hybrid individuals (Table 4). For animal 98/01
(the only reasonably certain F1 hybrid), the estimated
probabilities of ancestry from wild and domestic are
closest to 0.5:0.5, indicating that the fit by the model
Table 3. Population assignment using STRUCTURE analysis and
inferred ancestry of individual buffalo (a) that were predefined as wild
or domestic, but apparently misclassified, and (b) after the apparently
misclassified wild and domestic individuals were reassigned to the
hybrid group
Estimated ancestry
Population Wild Hybrid Domestic
(a)
Wild (n= 8)a0.971 0.024 0.005
27b0.809 0.074 0.117
69/01 0.863 0.132 0.005
Hybrid (n= 4)a0.014 0.985 0.001
32b0.011 0.725 0.264
35 0.248 0.739 0.013
73/01 0.194 0.788 0.017
Domestic (n= 21)a0.015 0.021 0.965
91/01b0.028 0.080 0.891
96/01 0.208 0.218 0.575
100/01 0.055 0.065 0.880
102/01 0.011 0.170 0.819
104/01 0.348 0.135 0.517
105/01 0.026 0.077 0.897
108/01 0.166 0.087 0.747
(b)
Wild (n= 8) 0.980 0.017 0.004
Hybrid (n= 16) 0.006 0.969 0.026
Domestic (n= 21) 0.014 0.030 0.956
aAverage inferred ancestry values for those individuals assigned to the same cluster as
that to which they were initially classified on the basis of phenotypic criteria
bAnimal identification number of apparently misclassified individuals
Table 4. Inferred ancestry of hybrid animals, given prior information
on the wild and domestic animals
Probability of ancestry
Wild Domestic
Wild (n= 8) 0.989 0.011
Domestic (n= 21) 0.018 0.982
Hybrids
27 0.372 0.628
69/01 0.727 0.273
91/01 0.494 0.506
96/01 0.725 0.275
100/01 0.532 0.468
102/01 0.426 0.574
104/01 0.677 0.323
105/01 0.397 0.603
108/01 0.627 0.373
31 0.832 0.168
32 0.272 0.728
33 0.700 0.300
35 0.698 0.302
73/01 0.651 0.349
98/01 0.598 0.402
99/01 0.700 0.300
is quite good. However, because of the small sample sizes
and the small genetic distances among the groups, 90%
probability intervals are very wide. Thus little weight can
be placed on the actual probabilities of ancestry, except to
say which of wild or domestic have contributed most to
each individual, e.g. domestic for animal 27, wild for
69/01, etc.
DISCUSSION
While it is known that matings do occur between wild
buffalo bulls and domestic cows (villagers releasing cows
into the forest), the hybrid progeny generally are returned
to the villages. Hybrid males that remain in the wild are
likely to be subordinate to wild males, and hence may not
breed, but hybrid females would be expected to breed with
wild males. To the extent that hybrid progeny of wild
males actually are returned to the villages, the gene pool
of the wild population will have suffered little
introgression from domestic buffalo. However, as some
animals originally classified as wild buffalo were
identified as being of mixed ancestry, some hybrids
have entered the wild population and phenotypic
criteria are not sufficient to define wild animals for a
conservation programme.
The method used for genetic admixture analysis and
population assignment assumes Hardy–Weinberg
equilibrium within populations and complete linkage
equilibrium between loci within populations. There were
no significant deviations from Hardy–Weinberg
equilibrium within our three predefined groups. Linkage
disequilibrium may be present, as some was detected in
our analysis of other buffalo populations. If so, it would
probably reduce the sensitivity (or accuracy) of our
analyses. However, deleting one locus involved in two
cases of linkage disequilibrium in other water buffalo
populations made no real difference to the assignments.
In addition, sample sizes (particularly for wild and hybrid)
were small. Nevertheless, in the STRUCTURE analysis
(Table 3(a)), two of the predefined wild animals (27 and
69/01) and seven of the predefined domestic animals
showed evidence of mixed ancestry. When these nine
animals were reclassified as hybrid, the fit of the model
was significant (all qi> 0.9 – Tables 3(b) and 4).
In selecting animals for a translocation programme, a
conservative approach would dictate that only those eight
animals identified here as wild should be included, at least
in the first instance and until more data become available.
The fact is that if a translocation programme is to be
attempted in order to establish a protected population,
more presumed wild animals will have to be sampled and
genotyped to determine their assignment. As their
genotypes are added to our existing database, assignments
would be expected to become more reliable, so that all
animals should be reassessed after each set of new animals
was genotyped.
Although the number of presumed wild animals that
could be sampled for this study was quite small (i.e. 10),
our results show that phenotypic criteria alone are not
sufficient for classification. Assignment tests, even with
such small numbers of putative wild individuals, can
provide a surer basis for conservation decisions.
Nevertheless, assignment test results are probabilistic, and
conservation management plans will need to consider
whether the aim is to attempt a genetically pure wild
population, or to accept some degree of introgression from
domestic buffalo. In the former case, the costs of obtaining
increased accuracy will have to be traded off against other
aspects of the programme (including effective size of the
founder population), but two approaches are possible:
(1) genotyping additional microsatellite loci; (2)
mitochondrial (mt) DNA sequencing. As introgression
from the domestic to the wild population will have been
primarily female mediated, mtDNA sequencing should
be particularly useful (Ward et al., 1999). Of 33 mtDNA
D-loop haplotypes found in 57 swamp and 23 river buffalo
(Lau et al., 1998), two of the three most common were
present in both swamp and river types, 23 (including the
third of the most common haplotypes) were found in
swamp only, and eight in river-only. Thus presence of a
river only haplotype in a presumed wild animal would
identify it as hybrid, while classification of an animal with
any other mtDNA haplotype would have to be based on
its microsatellite loci genotypes.
The primary data file (genotypes of all animals) is
available at http://ansc.une.edu.au/ansc/genetics/ link
through Barker, J. S. F.
Acknowledgements
This work was undertaken at the request of the
Department of National Parks and Wildlife Conservation,
Nepal, with a view to identifying a founder nucleus of
wild buffalo that could be moved from Kosi Tappu
Wildlife Reserve to the Royal Bardia National Park. The
work was funded by the Department for International
Development, United Kingdom and the Kadoorie
Charitable Foundations, Hong Kong. We are most grateful
to Dr J. T. Heinen, Florida International University, USA,
for assistance with phenotypic classification of animals,
and to Dr Bianca Moioli, Istituto Sperimentale per la
Zootecnia, Italy, and Professor H. Raadsma, University of
Sydney, for their gifts of microsatellite primers.
REFERENCES
Amano, T. (1983). Genetic differences between swamp and river
buffaloes in biochemical and immunological characteristics. In
Proceedings of the Preconference Symposium of the 5th World
Conference on Animal Production: 131–135. Shimizu, H. (Ed.).
Tsukuba, Japan.
Amano, T., Miyakoshi, Y., Takada, T., Kikkawa, Y. & Suzuki, H.
(1994). Genetic variants of ribosomal DNA and mitochondrial
DNA between swamp and river buffaloes. Anim. Genet. 25 (Suppl.
1): 29–36.
Barendse, W., Armitage, S. M., Kossarek, L. M., Shalom, A.,
Kirkpatrick, B. W., Ryan, A. M., Clayton, D., Li, L., Neibergs, H.
L., Zhang, N., Grosse, W. M., Weiss, J., Creighton, P., McCarthy,
F., Ron, M., Teale, A. J., Fries, R., McGraw, R. A., Moore, S. S.,
Georges, M., Soller, M., Womack, J. E. & Hetzel, D. J. S. (1994).
A genetic linkage map of the bovine genome. Nature Genet.
6: 227–235.
Barker, J. S. F., Moore, S. S., Hetzel, D. J. S., Evans, D., Tan, S. G.
& Byrne, K. (1997a). Genetic diversity of Asian water buffalo
(Bubalus bubalis): microsatellite variation and a comparison with
protein-coding loci. Anim. Genet. 28: 103–115.
Barker, J. S. F., Tan, S. G., Selvaraj, O. S. & Mukherjee, T. K.
(1997b). Genetic variation within and relationships among
populations of Asian water buffalo (Bubalus bubalis). Anim.
Genet. 28: 1–13.
Bowcock, A. M., Ruiz-Linares, A., Tomfohrde, J., Minch, E., Kidd,
J. R. & Cavalli-Sforza, L. L. (1994). High resolution of human
evolutionary trees with polymorphic microsatellites. Nature 368:
455–457.
Chen, Y. C. & Li, X. H. (1989). New evidence of the origin and
domestication of the Chinese swamp buffalo (Bubalus bubalis).
Buffalo J. 1: 51–55.
Cockrill, W. R. (Ed.) (1974). The husbandry and health of the
domestic buffalo. Rome: Food and Agriculture Organization of
the United Nations.
Di Berardino, D. & Iannuzzi, L. (1984). Detailed description of
RBA-banded chromosomes of river buffalo (Bubalus bubalis L.).
Génét. Sél. Evol. 16: 249–260.
Fischer, H. & Ulbrich, F. (1968). Chromosomes of the Murrah
buffalo and its crossbreds with the Asiatic swamp buffalo
(Bubalus bubalis). Z. Tierzücht. Züchtungsbiol. 84: 110–114.
Goudet, J. (1995). FSTAT (Version 1.2): a computer program to
calculate F-statistics. J. Hered. 86: 485–486.
Heinen, J. T. (2002). Phenotypic and behavioural characteristics
used to identify wild buffalo (Bubalus bubalis) from feral
backcrosses in Nepal. J. Bombay Nat. Hist Soc. 99: 173–183.
Heinen, J. T. & Singh, G. R. (2001). A census and some
management implications for wild buffalo (Bubalus bubalis) in
Nepal. Biol. Conserv. 101: 391–394.
Hochberg, Y. (1988). A sharper Bonferroni procedure for multiple
tests of significance. Biometrika 75: 800–802.
Lau, C. H., Drinkwater, R. D., Yusoff, K., Tan, S. G., Hetzel, D. J.
S. & Barker, J. S. F. (1998). Genetic diversity of Asian water
buffalo (Bubalus bubalis): mitochondrial DNA D-loop and
cytochrome b sequence variation. Anim. Genet. 29: 253–264.
Lessios, H. A. (1992). Testing electrophoretic data for agreement
with Hardy–Weinberg expectations. Marine Biol. 112: 517–523.
Macgregor, R. (1939). The domestic buffalo. Thesis presented to
the Royal College of Veterinary Surgeons (quoted by I. L. Mason
in Cockrill (1974: 1–47).
Mason, I. L. (1974). Species, types and breeds. In The husbandry
and health of the domestic buffalo: 1–47. Cockrill, W. R. (Ed.).
Rome: Food and Agriculture Organization of the United Nations.
Minch, E., Ruiz-Linares, A., Goldstein, D. B., Feldman, M. W. &
Cavalli-Sforza, L. L. (1995). Microsat (version 1.5e) : a computer
program for calculating various statistics on microsatellite allele
data. http://hpgl.stanford.edu/projects/microsat/programs/.
Moore, S. S., Byrne, K., Berger, K. T., Barendse, W., McCarthy,
F., Womack, J. E. & Hetzel, D. J. S. (1994). Characterization of
65 bovine microsatellites. Mammal. Genome 5: 84–90.
Nei, M. (1978). Estimation of average heterozygosity and genetic
distance from a small number of individuals. Genetics
89: 583–590.
Nei, M. (2002). Estimation of divergence times for a few
mammalian and several primate species. J. Hered. 93: 157–164.
Pritchard, J. K., Stephens, M. & Donnelly, P. (2000). Inference of
population structure using multilocus genotype data. Genetics
155: 945–959.
Raymond, M. & Rousset, F. (1995). GENEPOP (Version 1.2):
population genetics software for exact tests and ecumenicism. J.
Hered. 86: 248–249.
Saitou, N. & Nei, M. (1987). The neighbour-joining method: a new
method for reconstructing phylogenetic trees. Mol. Biol. Evol.
4: 406–425.
Scherf, B. D. (2000). World watch list for domestic animal diversity.
3nd edn. Rome: Food and Agriculture Organization of the United
Nations.
Takahata, N. & Nei, M. (1985). Gene genealogy and variance of
interpopulational nucleotide differences. Genetics 110: 325–344.
Tanaka, K., Solis, C. D., Masangkay, J. S., Maeda, K., Kawamoto,
Y. & Namikawa, T. (1996). Phylogenetic relationship among all
living species of the genus Bubalus based on DNA sequences of
the cytochrome b gene. Biochem. Genet. 34: 443–452.
Tanaka, K., Yamagata, T., Masangkay, J. S., Faruque, M. O., Vu-
Binh, D., Salundik, Mansjoer, S. S., Kawamoto, Y. & Namikawa,
T. (1995). Nucleotide diversity of mitochondrial DNAs between
the swamp and the river types of domestic water buffaloes,
Bubalus bubalis, based on restriction endonuclease cleavage
patterns. Biochem. Genet. 33: 137–148.
Ulbrich, F. & Fischer, H. (1967). The chromosomes of the Asiatic
buffalo (Bubalus bubalis) and the African buffalo (Cyncerus
caffer). Z. Tierzücht. Züchtungsbiol. 83: 219–223.
Ward, T. J., Bielawski, J. P., Davis, S. K., Templeton, J. W. & Derr,
J. N. (1999). Identification of domestic cattle hybrids in wild cattle
and bison species: a general approach using mtDNA markers and
the parametric bootstrap. Anim. Conserv. 2: 51–57.
Weir, B. S. & Cockerham, C. C. (1984). Estimating F-statistics for
the analysis of population structure. Evolution 38: 1358–1370.
Wright, S. (1951). The genetical structure of populations. Ann.
Eugenics (London) 15: 323–354.
Wright, S. (1978). Evolution and the genetics of populations. Vol.
4. Variability within and among natural populations. Chicago:
University of Chicago Press.