Phylogeographic Analysis of Tick-borne Pathogen - Recent Advances and Issues in Environmental Science - page 81

Caribbean, and South America, and then distances within and between geographical
clusters computed again.
Other analyses involved calculation of the percentage of changes between the
consensus MSP1a repeat sequences among different ecoregion clusters as a measure
of the genetic distance between A. marginale strains in these clusters. Some R1 and
RL sequences were found associated with more than one ecoregion cluster. There-
fore, the percentage of changes in amino acid composition for every sequence and
the consensus sequence in each cluster were computed as a measure of similarity of
each sequence and consensus sequence for each cluster. MSP1a repeat sequences were
aligned for pairwise comparison and determination of non-synonymous (d N ) and syn-
onymous (d S ) substitutions using Mega 4 [27]. The d N and d S were determined among
all pairwise comparisons of MSP1a repeat sequences within each ecoregion cluster,
estimated by the method of Nei and Gojobori [28] with the correction for multiple
substitutions [29]. The ratio of the mean d N /d S was used as an indicator of the level of
selection acting on MSP1a repeat sequences.
As an additional test to verify our hypothesis, we performed a Mantel's test. The
explanation of the distribution of the strains in terms of environmental variables may
be confounded because the variables are intercorrelated among themselves, and so it
may be diffi cult to ascribe causal mechanisms to the environmental variables. Mantels
test is a regression in which the variables are dissimilarity matrices. The operative
question is “do strains that have similar sequences also tend to be similar in terms of
the environmental variables”? Therefore, we performed a Mantel's test both between a
dependent distance matrix (genetic similarities of R1 and RL) and the predictor matrix
of geographical distances among strains. A second Mantel's test was done with the
same dependent matrix and a predictor dissimilarity matrix based on environmental
PCA-derived values. Such correlations will indicate if locations that are closer or loca-
tions that are similar environmentally are similar compositionally. Mantel's tests were
performed using the Jackard index according to [30].
Analysis of A. marginale MSP1a Microsatellite Sequences
The extent of genetic differentiation of A. marginale strains at the MSP1a microsat-
ellite was assessed within and among ecoregion clusters using an analysis of mo-
lecular variance (AMOVA; [31]) and pairwise population FST significance tests as
implemented in ARLEQUIN, version 3.01 [32]. The statistical significance of fixation
indices was tested using a non-parametric permutation approach [33] with 20,000 per-
mutations. Ecoregion clusters for which statistically significant subdivision was not
detected were pooled to define groups.
The effect of microsatellite size on MSP1a expression was characterized in A.
marginale strains Wetumka (OK; Genbank accession number AY010247), Okeechobee
(FL; AY010244), Idaho (ID; M32868) and HB-A8 (China; DQ811774) sequences.
These strains had microsatellite genotypes G (Wetumka and Okeechobee; distance
SD-ATG = 23 nucleotides), C (Idaho; SD-ATG = 19 nucleotides) and I (China; SD-
ATG = 29 nucleotides). The msp1alpha gene containing promoter sequences active in
Escherichia coli [6] was amplifi ed using oligonucleotide primers MSP1aP: 5'GCAT-