No abstract

Twenty-one microsatellite loci in 11 populations of Asian water buffalo (eight swamp, three river type) were analysed and, within and among populations, genetic variability was compared with results from 25 polymorphic protein-coding loci. Within-population mean heterozygosity ranged from 0.380-0.615, approximately twice that estimated from the protein-coding loci (0.184-0.346). Only eight significant departures from Hardy-Weinberg equilibrium (involving four loci) were detected; global tests showed significant heterozygote deficiencies for these four loci. Non-amplifying alleles are likely to be segregating in some or all populations for one of these loci, and probably for the other three. There was significant differentiation between the swamp and river types of water buffalo, and among populations within each buffalo type. Estimates of theta (measure of population differentiation) for each locus for the eight swamp populations were all highly significant (mean theta = 0.168 +/- 0.018). Mean theta for protein-coding loci was not significantly different (0.182 +/- 0.041). The variance among protein-coding loci was significantly higher than among microsatellite loci, suggesting balancing selection affecting allele frequencies at some protein-coding loci. Genetic distances show clear separation of the swamp and river types, which were estimated to have diverged at least 10,000-15,000 years ago. The topology of the swamp populations' microsatellite tree is consistent with their geographical distribution and their presumed spread through south-east Asia. By contrast, the tree based on the protein-coding loci distances is quite different, being clearly distorted by a bottleneck effect in one population, and possibly in at least two others. As many domestic livestock breeds are possibly descended from small numbers of founders, microsatellite-based trees are to be preferred in assessing breed genetic relationships.

Swamp and river buffalo mitochondrial DNA (mtDNA) was sequenced for 303 bp of the cytochrome b gene for 54 animals from 14 populations, and for 158 bp of the D-loop region for 80 animals from 11 populations. Only one cytochrome b haplotype was found in river buffalo. Of the four haplotypes identified in swamp buffalo, one found in all populations is apparently ancestral both to the other swamp haplotypes and to the river haplotype. The phylogenetic relationships among the 33 D-loop haplotypes, with a cluster of 11 found in swamp buffalo only, also support the evolution of domesticated swamp and river buffalo from an ancestral swamp-like animal, most likely represented today by the wild Asian buffalo (Bubalus arnee). The time of divergence of the swamp and river types, estimated from the D-loop data, is 28,000 to 87,000 years ago. We hypothesise that the species originated in mainland south-east Asia, and that it spread north to China and west to the Indian subcontinent, where the rive type evolved and was domesticated. Following domestication in China, the domesticated swamp buffalo spread through two separate routes, through Taiwan and the Philippines to the eastern islands of Borneo and Sulawesi, and south through mainland south-east Asia and then to the western islands of Indonesia.

Levels of variation in eight large captive populations of D. melanogaster (census sizes ∼ 5000) that had been in captivity for periods from 6 months to 23 years (8 to 365 generations) were estimated from allozyme heterozygosities, lethal frequencies, and inversion heterozygosities and phenotypic variances, additive genetic variances (VA), and heritabilities (h2) for sternopleural bristle numbers. Correlations between all measures of variation except lethal frequencies were high and significant. All measures of genetic variation declined with time in captivity, with those for average heterozygosities, VA, and h2 being significant. The effective population size (Ne) was estimated to be 185–253 in these populations, only 0.037–0.051 of census size (N). Levels of allozyme heterozygosities declined rapidly in two large captive populations founded from another wild stock, being reduced by 86% and 62% within 2.5 years in spite of being maintained at sizes of approximately 1000 and 3500. Estimates of Ne/N for these populations were only 0.016 and 0.004. Two estimates of Ne/N for captive populations of D. pseudoobscura from data in the literature were also low at 0.036 and 0.012. Consequently, the rate of loss of genetic variation in captive populations and endangered species may be more rapid than hitherto recognized. Merely maintaining captive populations at large census sizes may not be sufficient to maintain essential genetic variation.

The swamp type of the Asian water buffalo is assumed to have been domesticated by about 4000 years BP, following the introduction of rice cultivation. Previous localizations of the domestication site were based on mitochondrial DNA (mtDNA) variation within China, accounting only for the maternal lineage. We carried out a comprehensive sampling of China, Taiwan, Vietnam, Laos, Thailand, Nepal and Bangladesh and sequenced the mtDNA Cytochrome b gene and control region and the Y-chromosomal ZFY, SRY and DBY sequences. Swamp buffalo has a higher diversity of both maternal and paternal lineages than river buffalo, with also a remarkable contrast between a weak phylogeographic structure of river buffalo and a strong geographic differentiation of swamp buffalo. The highest diversity of the swamp buffalo maternal lineages was found in south China and north Indochina on both banks of the Mekong River, while the highest diversity in paternal lineages was in the China/Indochina border region. We propose that domestication in this region was later followed by introgressive capture of wild cows west of the Mekong. Migration to the north followed the Yangtze valley as well as a more eastern route, but also involved translocations of both cows and bulls over large distances with a minor influence of river buffaloes in recent decades. Bayesian analyses of various migration models also supported domestication in the China/Indochina border region. Coalescence analysis yielded consistent estimates for the expansion of the major swamp buffalo haplogroups with a credibility interval of 900 to 3900 years BP. The spatial differentiation of mtDNA and Y-chromosomal haplotype distributions indicates a lack of gene flow between established populations that is unprecedented in livestock.