We studied Rubus wahlbergii and related species in R. sect. Corylifolii, primarily in Sweden, using DNA microsatellites. The results show that the material can be divided into three species: R. wahlbergii s.s. which occurs along the Swedish coast from Bohuslän to Uppland, R. nordicus which occurs on the western coast all the way south to north-western Skåne, and a previously unrecognized taxon which we here describe as R. wahlbergioides. The latter has scattered localities in Halland, Skåne, Blekinge, Öland and Småland, but is common around Kivik in Skåne and Kalmar in Småland. It is characterised by leaves that are without felt underneath, terminal leaflets with a somewhat more triangular shape and coarser serration than those of R. wahlbergii and also has more glands on the stem and the inflorescence. In the northern part of the distribution, R. nordicus is easily distinguished morphologically from R. wahlbergii, e.g. by a strong tendency to form 7-foliolate leaves, but in Skåne and Halland, the two species are harder to distinguish. Still, genetically R. nordicus is more similar to R. sordirosanthus than to R. wahlbergii.

Samples (young leaves from the tip of the annual shoot) of R. wahlbergii and several related Rubusspecies were collected, mainly from Sweden, but also from Denmark, Norway and Germany. The localities are specified in Table S1 in the Supplementary Material. Vouchers are deposited in LD. The samples were stored in a freezer at –80˚C until DNA was extracted.

DNA was extracted with the 2× CTAB method (Doyle and Doyle, 1987). The concentration of the DNA was determined by a fluorometer and the samples were diluted with ddH₂O to ~14 ng/µL. Four pairs of microsatellite primers, described in Table S2, were used to analyse variable microsatellite loci. The three loci Ru105b, Ru117b and Ru275a were described in R. idaeus (Graham et al., 2004), whereas the locus mRaCIRRIV2A8 was described in R. alceifolius Poir. (Ansellem et al., 2001). They will here be referred to as 105b, 117b, 275a and 2A8, respectively. They have been successfully used also for other Rubus species (Stafne et al., 2005; Zhou, 2014; Hedrén et al., 2020; Ryde et al., 2021; Lewin et al., 2022). One primer in each pair was tagged with a fluorescing molecule (Cy5-) to enable the identification of the polymerase chain reaction (PCR) products in an automated sequencer (ALF Express II).

The PCR amplification consisted of 35 cycles of denaturation at 94 ℃ for 30 s, annealing at varying temperatures (see Table S2) for 30 s and polymerisation at 72 ℃ for 1 min (10 min in the last cycle). Each PCR was performed in a solution consisting of (in a total volume of ca. 6.4 µL): 4.4 µL ddH₂O, 0.66 µL 10´ reaction buffer (100 mM Tris-HCl, pH 8.3, 500 mM KCl and 15 mM MgCl₂), 0.56 µL dNTPs (2.5 mM for each nucleotide), 0.25 µL Cy5-tagged primer (10 µg/mL), 0.1 µL untagged complementary primer (25 µmol/mL), 0.03 µL Taq polymerase (5 u/µL; Applied Biosystems) and 0.6 µL template DNA (14 ng/µL).

The PCR products were mixed with formamide and appropriate size markers and were then heated to 94 ℃ for 25 min before they were separated by electrophoresis in a polyacrylamide gel. The DNA fragments were detected and analysed with the help of an ALF Express II analysis system. Samples that resulted in poor PCR reactions were purified with a Qiagen purification kit, after which the PCR and electrophoresis were redone. Samples that still resulted in poor amplification (unspecific amplification, dominance of short fragments etc.) were excluded.

The relative height of the amplified fragments on the resulting electropherogram was recorded on a five-graded scale for each sample to obtain an estimate of band intensity. Differences in relative band intensity between plants with overlapping banding patterns were matched with information on ploidy levels and interpreted as possibly due to variation in allele copy numbers. Similarly, differences in relative band intensity between plants of the same ploidy expressing different numbers of bands were also interpreted as due to variation in allele copy numbers. Several samples were re-amplified by PCR and rerun to verify the patterns of relative band intensity. Such analyses gave rise to closely similar differences in band intensity as the original runs, with only minor variation in peak height (typically, at most by a single unit along the five-graded scale used by us).

DNA-ploidy levels were estimated by flow cytometry for 22 samples. This analysis was performed by Plant Cytometry Services (http://www.plantcytometry.nl). Fresh leaf material was homogenised and mixed with an ice-cold buffer with DAPI (Arumuganathan and Earle, 1991). The amount of  DNA per cell nucleus was determined in a CyFlow ML with iceberg lettuce (Lactuca sativa) as an internal standard. The estimates were calibrated by using Rubussamples with known chromosome numbers. The DNA-ploidy levels are presented as fractional chromosome numbers from this calibration line between the ratio in fluorescence between the samples of known chromosome numbers and the internal standard.

The search for putative genetic relationships based on the microsatellite results was facilitated by the use of computer programs taking into consideration both the length of the various microsatellite alleles and the relative intensity of the PCR products (Lewin et al., 2022). Principal-coordinate (PCO) analyses were performed with the NTSYSpc v.2.1q software (Rohlf, 1994), using a pair-wise comparison of Jaccard coefficients (Jaccard, 1908) based on the occurrence of alleles for all four loci.