LFMM based p-values of each nucleotide for 115 ITS nucleotides were obtained after alignment and curation of data of these only 16 sites showed significant association with geographical variables (colored SNPs). These are SNPS NUMBER 11, 36, 41, 42, 51, 54, 56, 82, 100, 103, 195, 107, 109, 110, and 114.A similar analysis for trnH-psbA and trnL-F sequences also identified the nucleotides with potential adaptive value against geographical variables. The Manhattan plot of the LFMM analysis for trnH-psbA is provided in Fig. 5. Out of 86 nucleotides, about 12 SNPs are significantly associated with geographical variables. We obtained no significant association between trnL-F sequences and geographical variables.Fig. 5Manhattan plot of trnH- psbA sequences (The SNPs showing significant association (log p-value > 2) are indicated with their SNP number).These results indicate that at least some of the nucleotides within ITS and trnH-psbA genetic regions were under local selection and were adapted to geographical variables. To produce data on probable date of the sequences’ adaptation, we first carried out divergence time estimation based on merely adaptive nucleotides which were identified by LFMM analysis. This new divergence time estimate would be run against the all sequence data divergence time obtained. The difference in their time (Delta divergence time), may suggest the time past for sequences to be adapted.BEAST chronological tree of the studied taxa, based on merely adaptive ITS nucleotides, is provided in Fig. 6. As it is evident from this tree, the divergence time has been relatively older that all sequence data. That is, the mean time for total sequences was about 4 MY, while the same time for only adaptive nucleotides goes back to 6 MY. The Ln value of delta time may be related to the time past for adaptation of these sequences through local selective pressure.Therefore, if we consider the mean dt1 time (Divergence time for total sequences) = 4 MY, and the mean Dt2 time (Divergence time based on adaptive sequences) = 6, the Ln of delta time i.e. suggested adaptive time (At) = 0.7 MY. That means about 0.7 MY time took place for ITS sequences’ adaptation.Fig. 6BEAST chronological tree of adaptive ITS sequences.A similar analysis for trnH-psbA sequences produced the mean time for total sequences is about 1 MY, while the same time for only adaptive nucleotides reaches to 3.5 MY. The delta time = 2.5 MY, and therefore at = Ln (-2.5) = 0.9 MY(Fig. 7).This indicates that trnH-psbA sequences took relatively a longer time to be selected for adaptation, compared to that of ITS sequences in Avicennia species. This longer time may be related to the nature trnH-psbA chloroplast genome which is more variable and has a higher magnitude of mutations / nucleotide substitutions, compared to that of ITS sequences with a more conservative nature. Since we found no sequence in trnL-F region adapted to geographical variables, we did not carry out the above said analyses for these sequences.Fig. 7BEAST chronological tree of Avicennia species based on merely adaptive trnH- psbA sequences.Phylogenetic signal of adaptive sequencesPhylogenetic PCA (pPCA), of ITS data is provided in Fig. 8 of these adaptive nucleotides, SNP number 36, 41, 100, 109, and 114 shows a phylogenetic signal as they are present in A. alba, A. officinalis, and A. marina, which comprise a distinct phylogenetic clade.Fig. 8Phylogenetic PCA of ITS sequences.Moreover, some of these SNPs also shows close relationship between A. rumphiana and the above said clade. A similar analysis of trnH-psbA adaptive nucleotides (Fig. 9), revealed that the SNPs number 1, 3, 8, 9, 11, and 12 have some degree of phylogenetic signal as they show species relationship within a sub-clade of the tree.Fig. 9Phylogenetic PCA of Avicennia species based on trnH-psbA sequences.
