Historical processes of long-distance migration and ocean-wide expansion feature the global biogeographic pattern of Rhizophora species. Throughout the Indian Ocean, R. stylosa and R. mucronata appear as a young phylogenetic group with expansion of R. mucronata towards the Western Indian Ocean (WIO) driven by the South Equatorial Current. Nuclear microsatellites revealed genetic patterns and breaks, however, estimating propagule dispersal routes requires maternally inherited cytoplasmic markers. Here, we examine the phylogeography of 21 R. mucronata provenances across a >4,200 km coastal stretch in the WIO using R. stylosa as outgroup. Full length chloroplast genome (164,474 bp) and nuclear ribosomal RNA cistron (8,033 bp) sequences were assembled. Boundaries, junction point, sequence orientation and stretch between LSC/IRb/SSC/IRa/LSC showed no differences with the R. stylosa chloroplast genome. A total of 58 mutations in R. mucronata encompassing transitions/transversions, insertion-deletions and mononucleotide repeats revealed three major haplogroups. Haplonetwork, Bayesian ML and Approximate Bayesian Computation (ABC) analyses supported discrete historical migration events. An ancient haplogroup A in the Seychelles and eastern Madagascar was as divergent from other R. mucronata haplogroups as it was from R. stylosa. A star-like haplonetwork referred to recent range expansion of haplogroup B from northern Madagascar towards the African mainland coastline, including a single variant spanning >1,800 km across the Mozambique Channel Area. Populations south of Delagoa Bight contained haplogroup C and originate from a unique bottleneck dispersal event. Divergence estimates of pre- and post-Last Glacial Maximum illustrated a recent emergence of WIO Rhizophora mangroves compared to other oceans. Connectivity patterns could be aligned with directionality of major ocean currents. Madagascar and the Seychelles each harbored haplogroups A and B, albeit among spatially separated populations, explained from a different migration era. Likewise, the Aldabra Atoll harbored spatially distinct haplotypes. Nuclear ribosomal cistron (8,033bp) variants corresponded to haplogroups and confirmed admixtures in the Seychelles and Aldabra. These findings shed new light on the origins and dispersal routes of R. mucronata lineages that have shaped their contemporary populations in large regions of the WIO, which may be important information for defining marine conservation units, both at ocean scale and at level of small islands.

We considered Rhizophora samples from 21 mangrove populations distributed across the WIO (Table 1). Ten populations were located along a >4200 km stretch of the eastern African continent and eleven populations on islands in the WIO. Samples were from remote islands of the Seychelles (both sides of Mahé and Isle Curieuse) and Aldabra Atoll (three sites), from Madagascar (two sites on eastern coastline, two at northern tip and one along western coastline) and on the East African mainland coastline from Kenya, Mozambique and Republic of South Africa (ten sites). These were collected or provided and morphologically identified as R. mucronata following Tomlinson (2016) For each sample site, fresh leaves were collected from individual adult trees and stored in individual bags with silica gel for transport.

Genomic DNA extracts of 23 samples were made at the Plant Biology and Nature Management lab of the Vrije Universiteit Brussel (VUB) and processed for next generation sequencing analysis using the E.Z.N.A. SP plant DNA Mini Kit (Omega biotek, Norcross, GA, USA). After passing quality inspection (DNA concentration between 5 to 15ng/ul), the constructed library (TruSeq Nano DNA Kit) was sequenced by 300 bp x 2 paired-end sequencing in an Illumina MiSeq platform (Macrogen, Seoul, South Korea). Raw data was filtered out to remove the joint sequence and low-quality reads to obtain high-quality clean data. The Illumina pair-end next generation sequencing (NGS) product are used as the input file for de novo chloroplast assembles. The de novo chloroplast assemblies were done at first for three chloroplast genomes using NOVOPlasty assembly at Kmer = 33. All assemblies were executed by taking a single read from the dataset that originates from the targeted plastid as seed (rbcL) and taking 30% as subsample from the FASTA file with default parameters. These were compared to the existing annotated R. stylosa chloroplast genome (GenBank accession number MK070169 and appeared similar in genome structure and perfectly aligned with R. stylosa though did not align well with GenBank accession number MN307165 (Rhizophora mucronata, Wu, H. unpublished) or MW387538 (Rhizophora apiculata, Jiang, G.-F. unpublished). Therefore, the remaining samples were assembled using ‘assemble to reference’ function in Geneious software. Illumina 2 x 300 bp paired-end were processed in Geneious Prime v 2019.2.1 (© 2005-2019 Biomatters Ltd.) to obtain complete chloroplast genome sequences. All assemblies used MK070169 (R. stylosa) as a reference genome and the mapping of reads of 23 samples were performed. This approach yielded identical results as obtained with Novoplasty and the assemblages using R. stylosa as a reference to map every sample, averaged 66,892 – 454,099 reads with a mean depth of reads ranging from 118 – 829 coverage.