Nucleospin 96 plant kit

We sampled silica gel-dried leaves or cambium from trees during multiple field expeditions in nine countries, covering most of the natural distributions of the two Lophira species (Fig. 1a). One of these expeditions focused on the forestsavanna mosaic area of Cameroon where the two species occur in close vicinity. Another expedition, in Gabon, sampled trees near the town of Ndjolé where a contact zone between two genetic groups of L. alata was identified during preliminary data analyses. Total genomic DNA from 433 L. alata and 370 L. lanceolata individuals was isolated using the NucleoSpin 96 plant kit (Macherey-Nagel, Düren, Germany) according to the manufacturer's protocol. We amplified 13 polymorphic nuclear microsatellite loci (namely P12, P18, P24, P31, P34, P36, P40, P44, P47, P51, P53, P62, P66) isolated from L. alata and optimized in two multiplex reactions according to Piñeiro et al. (2015) (link). Three loci (P24, P36, P44) that suffered from weak amplification were discarded. When an individual presented no trace of PCR amplification product for a given locus while alleles were clearly visible at the other loci amplified in the same PCR mix, it was considered homozygote for a null allele.

DNA was extracted with the Nucleospin® 96 plant kit (Macherey-Nagel, Düren, Germany), from 3-38 mg of dried tissue, following manufacturer's instructions, but with a modified lysis step at room temperature for one hour instead of 65°C for 20 min. Extracted DNA was diluted to 1:100. We chose to use microsatellite markers as a cost-effective technique to examine a large number of individuals. These markers have been shown to provide similar results to ddRad-Sequencing for diversity and connectivity analyses of U.
pinnatifida in the study regions (Guzinski et al. 2018) . As compared to ddRad-Sequencing, the use of microsatellites makes also easier the comparison with our previous dataset (Guzinski et al. 2018 , Grulois et al. 2011) . A total of 816 individuals were genotyped with 10 microsatellite loci, Up-AC-1B2, Up-AC-1B5, Up-AC-1C1, Up-AC-1G2, Up-AC-1H5, Up-AC-2C1, Up-AC-2E8, Up-AC-4G2, Up-AC-4C12, Up-AC-4E9 (Daguin et al. 2005 ) using multiplex PCRs, as described in Grulois et al. (2011) . PCR products were run with a capillary sequencer AB3130XL (Applied Biosystem). Genotypes were scored with GeneMapper 4.0 (Applied Biosystem). Missing data represented 2.8% of all data and were distributed across loci and samples.

Individuals of L. rodriguezii were sampled from four sites in the Mediterranean, three in Eastern Provence (Banc Magaud_1, Banc Magaud_2, Cap Camarat) and one in Southern Corsica (Bonifacio), between 65 to 76 meters depth (Figure 1). Sampling was done by scuba divers between June and August 2018. Sampling for DNA collection was performed by collecting a small piece of tissue from the blade of sporophytes. More precisely, a total of 47 samples were collected from sporophytes not connected to each other by a stolon, and with a minimum distance of two meters between sampled blades. This strategy aimed to promote the sampling of distinct genets. For one of the two Banc du Magaud sites, namely Banc Magaud_2, we used a different sampling design: the spatial distance between the sampled sporophytes was recorded along a regular transect of 20 meters to study the spatial genetic structure inside the population. Regarding L. digitata, we used samples from five northeastern Atlantic populations of L. digitata collected in 2018 (Table S1, Supporting Information).
Genomic DNA was extracted using the Nucleospin® 96 plant kit (Macherey-Nagel, Düren, Germany), according to the manufacturer's protocol.