The present study includes two aspects in the field of phylogeography. First, the technology of High Resolution Melting (HRM) analysis based on real-time PCR is introduced as a tool for haplotype detection in phylogeographic research. Second, phylogeographic study of three arctic-alpine or alpine plant species in the family Caryophyllaceae is carried out based on the haplotype data obtained through established protocol of HRM analysis. In Chapter 2, experimental protocols of HRM analysis combined with posterior DNA sequencing as a complementary and confirmatory method are established for haplotype detection in the case study of three species, Arenaria ciliata, A. norvegica and Minuartia recurva. Non-coding chloroplast DNA loci, rps16 intron (c.750bp for Arenaria and c.690bp for Minuartia) and trnT-trnL (c.640bp for Arenaria and c.540bp for Minuartia) are used in HRM analysis, where they are split into smaller (<400bp) amplicons for each real-time PCR reaction. The protocol is able to reveal 19 out of 20 haplotypes of rps16 and all of the 24 haplotypes of trnT-trnL in the case of Arenaria species, and to reveal all of the four haplotypes of rps16 and three out of eight (or five if variation in SSRs is not considered) haplotypes of trnT-trnL in the case of M. recurva. Posterior DNA sequencing reveals only one more haplotype with rps16 in the case of Arenaria species, which indicates a high sensitivity of HRM analysis in both cases. In Chapter 3 and 4, phylogeographic studies are carried out for the Arenaria species and M. recurva. Based on the haplotype identities and their distribution within and among the sampled populations, a complete phylogeographic study becomes possible. Maximum-likelihood phylogenetic trees and statistical parsimony networks are constructed among the haplotypes, and the genealogical relationship among the haplotypes is combined with their geographic distribution to understand the migratory history of the populations, population genetic analysis is made to understand the genetic diversity and structure of the populations and mismatch analysis is performed to understand demographic history of the populations. In Chapter 3, deeply diverged clades are revealed which are not in accordance of either subspecies or geographic localities. The data indicate a much older establishment of the populations of A. ciliata on Ben Bulben in northwest Ireland than thought before, possibly as early as 150-250 thousand years ago. Thus these Irish populations may have survived the last ice age in situ, rather than having immigrated after the end of Pleistocene (c. 12, 000 years ago). However, the result shows that the Irish populations are more closely related to the Iberian populations than they are related to the Alps populations. In Chapter 4, much lower level of genetic polymorphisms is revealed in the M. recurva populations, although they cover a European range comparable to that in the Arenaria case. The Balkans region is suggested as the refugium for the species, while little variation is found across the species distribution from the Alps, the Pyrenees, the north side of Spain and Ireland, which indicates a recent dispersal of the species in west Europe. Also a close relationship is suggested between the Irish populations and the Iberian populations. In Chapter 5, a further theoretical assessment of possibility of missed detection in HRM analysis is carried out via in silico HRM simulation. Based on an amplicon of rps16I in A. ciliata, random mutations are made and HRM sensitivity is evaluated to different classes of single nucleotide substitutions. Class I and II class I (A/G or C/T) and II (A/C or G/T) substitutions are demonstrated easier to be detected than class III (C/G) and IV (A/T) substitutions. Further analyses suggest that between 50 and 650bp, amplicons of greater sizes are more likely to yield multiple melting peaks, which is favourable for higher sensitivity in HRM analysis. Between 100 and 550bp when all the amplicons render double melting peaks, amplicons of greater sizes tend to provide lower sensitivity in HRM analysis. Amplicons smaller than 350bp with double melting peaks are considered to generate an acceptable rate of missed detection (<10-20%). In addition, in silico HRM analysis is tested with available DNA sequences of mitochondrial and chloroplast loci from published phylogeographic studies, and is demonstrated to help distinguish most of the extant haplotypes, although with a proportion of haplotypes missed (typically 10-20%). The results provide information for possible improvements of HRM analysis to be widely applied for haplotype detection in phylogeographic research.