Resolving the identity of commercially cultivated Ulva (Ulvaceae, Chlorophyta) in integrated seaweed‐abalone aquaculture farms in South Africa

Species of Ulva have a wide range of commercial applications and are increasingly being recognized as promising candidates for integrated aquaculture. In South Africa, Ulva has been commercially cultivated in integrated seaweed‐abalone aquaculture farms since 2002, with more than 2000 tonnes of biomass cultivated per annum in land‐based paddle raceways. However, the identity of the species of Ulva grown on these farms remains uncertain. We therefore characterized samples of Ulva cultivated in five integrated multi‐trophic aquaculture farms (IMTA) across a wide geographical range and compared them with foliose Ulva specimens from neighboring seashores. The molecular markers employed for this study were the chloroplast‐encoded Ribulose‐1,5‐bisphosphate carboxylase oxygenase (rbcL), the Internal Transcribed Spacer (ITS) of the nuclear, and the chloroplast elongation factor tufA. All currently cultivated specimens of Ulva were molecularly resolved as a single species, U. lacinulata. The same species has been cultivated for over a decade, although a few specimens of two other species were also present in early South African IMTA systems. The name Ulva uncialis is adopted for the Ulva “Species A” by Fort et al. (2021), Molecular Ecology Resources, 22, 86) significantly extending the distribution range for this species. A comparison with wild Ulva on seashores close to the farms resulted in five new distribution records for South Africa (U. lacinulata, U. ohnoi, U. australis, U. stenophylloides, and U. aragoënsis), the first report of a foliose form of U. compressa in the region, and one new distribution record for Namibia (U. australis). This study reiterates the need for DNA confirmation, especially when identifying morphologically simple macroalgae with potential commercial applications.


INTRODUCTION
Ulva, also known as sea lettuce, is a genus in the Chlorophyta with a wide range of commercial applications and is a promising candidate for integrated aquaculture.The commercial cultivation of Ulva in integrated seaweed-abalone aquaculture farms began in South Africa in 2002.Free-floating wild populations of Ulva were initially collected from sheltered waters (harbors or bays) close to the abalone farms and were integrated into existing aquaculture systems (Bolton et al., 2009;Troell et al., 2006).Ulva strains from South Africa undergo somatic growth in culture and, therefore, do not become fertile, which limits the need to control their life histories (Bolton et al., 2009;Troell et al., 2006).A recent study (wrongly) reported that Ulva could only be cultivated from spores (Balar & Mantri, 2020), but this is not generally the case in land-based Ulva systems.In South Africa, fertility has not been observed in culture, and no bladed Ulva grows on the sides of the paddle raceway systems, further evidence of a lack of spore production.It has been known for many years that certain strains of Ulva are beneficial for vegetative production in landbased systems.For example, in pioneering studies of land-based Ulva cultivation, Ryther (1982) stated: "Dr.Howard Levine (U.Mass.) kindly provided the author with a sample of Ulva lactuca from a population which he had never observed to become reproductive or to bear fruiting bodies.In the year since it has been grown in our culture system, it has also remained sterile, in contrast to several other strains of the same species grown under the same conditions " (p. 19).
More than 2000 t of Ulva are grown per annum in South Africa in large integrated oval paddle raceways (~30 m long, 0.5-1 m deep), mostly in abalone effluent (Neveux et al., 2018).The cultivated biomass is mainly used as supplementary feed for the abalone and for the bioremediation of effluent water (Bolton et al., 2013(Bolton et al., , 2016;;Neveux et al., 2018).The bioremediation capacity of Ulva allows for partial recirculation of water on the farm, which reduces pumping costs and increases the water temperature in these land-based paddle raceways, which in turn leads to faster abalone growth (Bolton et al., 2009;Nobre et al., 2010).
The main integrated seaweed-abalone aquaculture farms in South Africa are located over ~1925 km of coastline ranging from the cool temperate northern west coast of South Africa (Diamond Coast Aquaculture) to the warm temperate southeast coast (Wild Coast Abalone).The very different ambient seawater temperature regimes at these farms range from 17.9 to 18.6°C (monthly mean temperature) at Wild Coast Abalone in Haga-Haga to 11.3-13.3°Cat Diamond Coast Aquaculture Farm in Kleinsee (Smit et al., 2017).It is therefore hypothesized that farms over such a large biogeographic range, encompassing different marine biogeographical provinces (Anderson et al., 2009) may be expected to grow different species of Ulva.
The taxonomy and subsequent identification of species of Ulva is difficult due to a substantial discordance between morphological and molecular identification (Tran et al., 2022).As such, species misidentifications in the genus are not uncommon, many species are referred to under different names in different regions of the world, and different species sometimes bear the same name (Fort et al., 2021;Tran et al., 2022).A few major studies have been carried out recently on Ulva molecular species determination and barcoding, and there is a growing consensus on matching species names with clades using specific molecular markers (e.g., Fort et al., 2021;Loughnane et al., 2008;O'Kelly et al., 2010).The nomenclature of Ulva species, the names given to these molecular clades, has therefore been changing considerably, both from the expansion of molecular investigations and from recent studies that have sequenced various type specimens of Ulva (e.g., Hughey et al., 2019Hughey et al., , 2022)).For instance, the name U. lactuca has been misapplied in many studies around the globe, and 65% of all entries of U. lactuca in the NCBI database typically represent incorrectly labeled specimens of other species of Ulva (Fort et al., 2021;Hughey et al., 2019).Similarly, high-throughput sequencing of the type specimen of U. rigida placed it in a separate clade from specimens erroneously identified as the widely distributed U. rigida.Instead, the clade containing the lectotype of U. rigida and specimens previously misidentified as U. rotundata from Portugal and Ireland, which therefore represent the true U. rigida, were shown to be confined to European waters (Hughey et al., 2022).The clade containing the previously misidentified specimens of the widely distributed U. rigida also contained the lectotype of U. lacinulata from its type locality in Lesina, Croatia, along with the type material of U. armoricana, U. laetevirens, and U. scandinavica (Hughey et al., 2022).Since U. lacinulata is the oldest validly published name in this clade, it is the correct name that should be applied to the globally distributed species that was previously known as U. rigida (Hughey et al., 2022).However, according to Fort et al. (2021), U. lacinulata and U. rigida are two separate species, and the sequences previously assigned to U. rigida are provisionally referred to as Ulva sp.A by Fort et al. (2021).Since there are currently no described species that are synonymous with the Ulva sp.A specimen in public repositories, the sequences of Ulva sp.A will need to be renamed when a suitable valid described species is resolved within this clade (Fort et al., 2021).
Ulva that have been commercially grown in integrated aquaculture farms in South Africa have similarly been given different species names in the literature in the past few years based on morphological identification, and identification is especially difficult when these free-living specimens have no basal portion (Cyrus et al., 2015;Robertson-Andersson et al., 2008;Shuuluka et al., 2013).The first attempt to confirm the identity of this cultivated Ulva used the DNA marker ITS region, together with morphology, to identify samples collected from seashores and from one aquaculture farm, Irvin & Johnson Cape Abalone (Kandjengo, 2002).Based on morpho-anatomical and cytological characters, four species of Ulva-U.rigida, U. capensis, U. lactuca, and U. linza-were identified from the Irvin & Johnson Cape Abalone Farm.However, all four morphological species were resolved as a single molecular species, considered then as U. rigida (sensu Stegenga et al., 1997;Kandjengo, 2002).The molecular data suggested that U. capensis and U. rigida from South Africa represent a single polymorphic species and included a sample collected from the Irvin & Johnson Cape Abalone farm.
Only a few studies on the taxonomy of Ulva have been carried out in South Africa.According to Joska (1992), the name U. uncialis was considered synonymous with U. rigida on morphological grounds.According to Silva et al. (1996), U. capensis should be known as U. uncialis because Phycoseris uncialis is an older name and was considered synonymous with U. capensis by Papenfuss.Since Kützing (1849, p. 475) had already proposed a new species for Drège's material (P.uncialis), U. capensis must be considered superfluous and hence illegitimate.Following Joska (1992), a comprehensive morphological documentation of seaweeds of the South African west coast was conducted by Stegenga et al. (1997).Within the genus Enteromorpha (tubular species), six species were identified, namely, E. intestinalis, E. atroviridis, E. linza, E. compressa, E. prolifera, and E. flexuosa (Stegenga et al., 1997; all these species are now in the genus Ulva).In the genus Ulva (foliose species), five species were identified, namely, U. fasciata, U. capensis, U. rigida, U. rhacodes, and U. lactuca (Stegenga et al., 1997).
The uncertainty associated with morphological identification coupled with the growing nomenclatural changes in Ulva over the last few years led to the aim of this study, which was to identify the species of Ulva grown in the main integrated seaweed-abalone farms in South Africa.Five farms-Diamond Coast Aquaculture, Abagold, Irvin & Johnson (I&J) Cape Abalone, Buffeljags Abalone, and Wild Coast Abalone-that are responsible for the production of a large majority of the cultured Ulva in South Africa were sampled.Foliose Ulva specimens from nearby rocky shores close to the farms in the Hermanus abalone farming complex in New Harbor and available sequences from a previous study (Kandjengo, 2002) were included in our analyses for comparative purposes.We were specifically interested in determining whether the farmed species were also present in wild marine habitats close to the farms.Specimens of Ulva were identified morphologically, and three molecular markers, rbcL, ITS rDNA region, and tufA, were used to confirm their identities.This paper will thus elucidate the identity of important commercial Ulva material and is the first study to place selected Ulva samples from Southern Africa into the recently developing consensus on Ulva molecular species identification.

Collection sites
Fresh Ulva specimens were collected from five abalone farms, Irvin & Johnson Cape Abalone, Abagold, Buffeljags Abalone, Diamond Coast Aquaculture, and Wild Coast Abalone.Fresh Ulva blades (a total of 12 specimens) were collected from land-based paddle-raceway systems, and attached Ulva specimens (four samples close to the farm inlet pipes-joint water intake of the farms-and 3 samples from near the outlets-outflow pipes of the farms) observed near the main abalone farming area in Hermanus were also collected.An additional 14 specimens were collected from nearby seashores (Kommetjie, Sea Point, Eersterivier, and Knysna) as well as foliose Ulva from Namibia (Swakopmund and Lüderitz), made available through the intertidal biodiversity survey conducted by the Benguela Current Commission (BCC) in 2019 (Figure 1; Table 1).Specimens were identified morphologically using the detailed descriptions in Stegenga et al. (1997).The size, texture, and color of the Ulva specimens were assessed visually, and the maximum length and width of the blades were measured.The anatomical characters recorded included the dentations on the blade margin, blade thickness, vegetative cell lengths and cell width, number of pyrenoids per cell, shape of cells in surface view, and the chloroplast arrangement in the cells (Bachoo, 2021).Photomicrographs were taken using an Olympus D50 digital camera mounted on the microscope, and a scale bar was included (Appendix S1 in the Supporting Information).All specimens were pressed and dried on herbarium sheets for deposit in the Bolus Herbarium at the University of Cape Town (BOL) or strorage in the Seaweed Biorepository at the University of Cape Town.Fresh Ulva samples were stored in prelabeled plastic bags and immediately stored in a −80°C freezer for DNA analysis.Additional sequences from three farms and nearby seashores (ITS rDNA region: 14 sequences; rbcL: 14 sequences) were added from a previous unpublished study (Table 1).

DNA extraction and PCR amplification
DNA was extracted from fresh or silica-dried Ulva samples using the QIAamp DNA Micro Kit (50; Cat# 56304) following the manufacturer's instructions with minor modifications that included the elution of DNA in 30 μL of elution buffer and incubation with the elution buffer for 5 min at room temperature before centrifugation.Samples were ground in liquid nitrogen, and ≤10 mg was used for DNA extraction.Prior to PCR amplification, the quantity and quality of the extracted DNA was estimated using a Nanodrop spectrophotometer (GEN-OVA NANO, JENWAY, Bibby Scientific Ltd.), and the integrity of the extracted DNA was assessed on a 1% agarose gel to ensure the presence of high molecular weight DNA.Fragments of the rbcL, ITS rDNA region, and tufA gene regions were PCR amplified using a Bio-Rad C1000™ thermal cycler with the primers RH1 (5′-ATGTC ACC ACA AAC AGA AAC TAAAGC-3′) and 1385r (5′-AATTC AAA TTT AAT TTC TTTCC-3′; Manhart, 1994), primers ITS5 (5'-GGAAG TAA AAG TCG TAA CAAGG-3′) and ITS4 (5′-TCCTC CGC TTA TTG ATATGC-3′; White et al., 1990), and primers tufGF4 (5′-GGNGC NGC NCA AAT GGAYGG-3′; Saunders & Kucera, 2010) and tufAR (5′ CCTTC NCG AAT MGC RAA WCGC-3′; Famà et al., 2002), respectively.All PCR reaction mixtures (25 μL volume) were prepared using 2 μL of approximately 50 ng • μL −1 DNA extract, 1× (12.5 μL) PCR-Master Mix (KAPA Taq ReadyMix, Kapa Biosystems; Catalog #KK1006), 0.5 μL (400 nM) of each primer, and 9.5 μL of sterile water.The PCR amplification thermal profile for each marker is listed in Table S1 in the Supporting Information.Polymerase chain reaction products were assessed on a 1% agarose gel to verify reaction specificity and fragment size, and PCR products were purified (Roche) and sequenced at the Central Analytical Unit at Stellenbosch University using a BigDye Terminator Cycle Sequencing Kit (Applied Biosystems) and AB13730xl Genetic Analyzer (Applied Biosystems) according to the manufacturer's instructions.Some PCR products were sent to Macrogen-Europe for sequencing and were purified by centrifugation using the Wizard® SV Gel and the PCR Clean-Up System prior to the sequencing reactions.Both forward and reverse primers for each gene were used for cycle sequencing.

Phylogenetic analyses
A total of 33 rbcL gene (1501 bp), 20 ITS rDNA region (1316 bp) and 33 tufA gene (1224 bp) sequences were obtained in the present study, and 14 rbcL gene and 14 ITS rDNA sequences were included from a previous unpublished study (L.Kandjengo, unpub. data).Additional Ulva sequences were selected from the nucleotide collection available in GenBank and relevant sequences from previous publications were added for comparison with Southern African material (Fort et al., 2021;Hughey et al., 2022;Kirkendale et al., 2013;Loughnane et al., 2008;Melton et al., 2016).This yielded multiple alignments with a total of 76 rbcL gene, 82 ITS rDNA region, 83 tufA gene sequences and a concatenated alignment of 60 rbcL and tufA gene sequences.All sequences were edited using the software BioEdit Sequence Alignment Editor version 7.0.5.3 (Hall, 1999).Phylogenetic analyses for the rbcL gene, ITS rDNA region, tufA gene sequences, and concatenated datasets were performed using the maximum likelihood (ML) analysis in RaxML via the online portal CIPRES (Miller et al., 2010) and Bayesian inference (BI) analysis in MrBayes v3.2 (Ronquist  Huelsenbeck, 2003).For the Bayesian analysis, two concurrent Markov chain Monte Carlo (MCMC) runs, each composed of four chains-three heated and one cold-were performed.Each Markov chain was run for 5,000,000 generations for the rbcL gene, ITS rDNA region, tufA gene, and concatenated datasets, sampling trees every 1000 generations.The first 25% of trees were discarded as burn-in.Parameter stability and run convergence were inspected using Tracer v1.7.1 (Rambaut et al., 2018).The consensus tree was generated from the remaining 75% of trees.For the maximum likelihood trees, the best evolutionary model for each of the three alignments was determined based on their Akaike Information Criterion (AIC) score calculated in jModeltest 2 (Darriba et al., 2012;Guindon & Gascuel, 2003).The TIM3, TIM1, and GTR + I + G were the most appropriate models for the rbcL gene, ITS rDNA region, tufA gene, and concatenated alignments, respectively.Maximum likelihood trees were obtained using RaxML-NG, and the trees were run for 100 bootstrap replicates (Kozlov et al., 2019).All trees were visualized using Figtree (http://tree.bio.ed.ac.uk/) and edited using Corel Draw.

RESULTS
Specimens of Ulva (n = 12) collected from land-based paddle raceway systems from various integrated seaweed-abalone aquaculture farms in South Africa were morphologically identified as either U. rigida or U. lactuca (Bachoo, 2021).However, all specimens were resolved in a single DNA clade containing the type specimen of U. lacinulata with low support (posterior probabilities, PP: 0.60) for the rbcL gene (Figure S1 in the Supporting Information) but moderate support for tufA gene (Figure S2 in the Supporting Information), ITS rDNA region (Figure 2), and the concatenated dataset (ITS-PP: 0.92; tufA-PP: 0.97 & bootstrap percentage,BP: 92; Concat-PP: 0.92 & BP: 92; Figure 3).A single sample (AG-IN3), attached to farm infrastructure collected near the inlet pipes of the Hermanus farming complex, was morphologically identified as U. rigida, but resolved in the same clade.Specimens of Ulva (n = 16) similarly collected from land-based paddle raceways from three farms, Irvin & Johnson Cape Abalone, Abagold, and Wild Coast Abalone in 2009 were morphologically identified as U. rigida, U. capensis, U. linza, and U. lactuca, but were all resolved within the larger U. lacinulata clade with low support (PP: 0.60) for the rbcL gene (Figure S1) and high support (PP: 0.92) for the ITS rDNA region (Figure 2).
In our concatenated phylogeny, Ulva lacinulata (PP: 0.97, BB: 92) and U. sp.A (PP: 1, BB: 100) were recovered in distinct sister clades (Figure 3).The latter clade contained specimens morphologically identified as U. uncialis.For all other genes, U. uncialis was recovered as an internal clade within the larger U. lacinulata with The topology was congruent with the ML tree.The tree was rooted with Percursaria percursa, Umbraulva olivascens, and Ulvaria obscura.
Posterior probability values (PP) for the Bayesian analysis followed bootstrap values for the maximum likelihood (ML) analysis (≥70) and are indicated at each node.South African specimens are in gray.Numbers accompanying the species names are GenBank accession numbers for the sequences used in the analysis.
moderate to high support (rbcL-PP: 0.7; tufA-PP: 1, BP: 97; ITS-PP: 0.75, BP: 70).Four farmed specimens (U03, U16, U21, U21F) collected in 2009 were also resolved in the internal U. unciali clade for the ITS rDNA region (Figure 2) and rbcL gene (Figure S1), and three specimens (U09, U134 and U135) were identified as U. stenophylloides for the ITS rDNA region (Figure 1).However, no farmed material after 2009 contained specimens of U. unciali or U. stenophylloides, although seashore material and specimens attached to farm infrastructure just outside the farms was consistent with U. uncialis (D3049 and AG-OUT1).Specimens from nearby seashores or in close proximity to the aquaculture farms were morphologically identified as Ulva rigida, U. capensis, U. lactuca, U. fasciata, U. uncialis, and U. flexuosa, but all were resolved as different molecular species with high support (Figure 3).Similarly, specimens from nearby seashores collected in 2009 were morphologically identified as U. rigida, U. capensis, U. uncialis, and U. lactuca but were resolved as different molecular species with high support (Figure 3).One specimen (U03) collected on the seashore from Yzerfontein on the west coast of South Africa in 2009 resolved within the U. lacinulata DNA clade, which included the cultivated Ulva specimens.The molecular identification of seashore specimens included five new distribution records-U.ohnoi, U. australis, U. stenophylloides, U. lacinulata, and U. aragoënsis-and the first record of a foliose form of U. compressa for South Africa and one new distribution record for Namibia (U. australis).

D ISCUSSION
Our results indicate that all currently cultivated Ulva in integrated seaweed-abalone aquaculture farms in F I G U R E 3 Bayesian Inference tree of Ulva species from the concatenated dataset, tufA and rbcL genes.The tree was rooted with Ulvaria obscura.Posterior probability values (PP) for the Bayesian analysis followed bootstrap values for the maximum likelihood (ML) analysis (≥70) and are indicated at each node.South African specimens are in gray.The asterisks (*) next to the five Ulva species indicate new Ulva records for Southern Africa.
South Africa are the same species, despite wide geographical distribution and contrasting environmental conditions.All locally farmed Ulva specimens were identified as U. lacinulata following the recent nomenclatural change by Hughey et al. (2022).Ulva lacinulata is the oldest and therefore the valid taxonomic name for the molecular clade previously identified as U. laetevirens sensu Fort et al. (2021) that contained the type specimens of various species including U. armoricana.Ulva armoricana is a name that has been used for Ulva cultivated in South Africa (Cyrus et al., 2015).Ulva laetevirens was recently synonymized with U. australis by Hughey et al. (2021), and the lectotype specimen of U. lacinulata was resolved within the clade containing U. laetevirens sensu Fort et al. (2021) by Hughey et al. (2022), which means that the new valid name for this clade is U. lacinulata.
In their attempt to match names to DNA clades, Hughey et al. (2022) recognized a single, wellsupported rbcL clade, to which they assigned the name Ulva lacinulata.This clade, however, contained sequences that were recovered in two separate clades according to Fort et al. (2021).Ulva lacinulata and Ulva sp.A, a closely related and undescribed species, were recovered as distinct clades based on a comprehensive collection of tufA gene, ITS1 rDNA region, and rbcL gene sequences to which different species DNA delimitation methods were applied (Fort et al., 2021).At the time, Ulva sp.A contained specimens misidentified as U. rigida or U. lactuca, but there were no available species names that could be unambiguously assigned to it.In the present study, specimens consistent with the morphology of U. uncialis (previously U. capensis) were firmly placed within the Ulva sp.A clade for all individual gene phylogenies and for the well-supported concatenated phylogeny.We therefore propose using the name U. uncialis for the clade previously referred to as Ulva sp.A by Fort et al. (2021).However, future studies and sequencing of the type specimen of Phycoseris uncialis, the basionym of the species, is still needed to confirm our assignment of the clade.
Molecular support for the Ulva lacinulata and U. uncialis clades was generally higher for the concatenated phylogeny and for the fast evolving ITS rDNA region and the tufA gene, but lacked resolution for the rbcL gene which might explain the lack of distinction within the U. lacinulata rcbL clade by Hughey et al. (2022).Other studies have similarly found that tufA is the most suitable molecular marker for species-level identification of Ulva species compared with the other commonly used markers, rbcL gene and ITS rDNA region.Although it has better amplification success and higher resolution compared with the other markers, the tufA gene is limited by a lack of available reference sequences for comparison, which makes linking Linnaean names to species clades particularly difficult (Tran et al., 2022).Nevertheless, the lack of a reference library for the tufA gene should not limit its use in future studies.Our combined phylogenetic analysis of tufA and rbcL genes shows even better resolution than any single gene tree and is therefore recommended for future studies of this nature.Despite the low rbcL divergence between U. lacinulata and U. uncialis, we believe that these two entities should be tentatively considered different species as South African specimens are easily distinguished based on morphological characters.Ulva lacinulata has a thick, dark thallus, smooth margin and rectangular cells, and U. uncialis has a lighter, thinner thallus, teeth along the thallus margin and bullet shaped cells (see the description of Ulva capensis in Stegenga et al., 1997).However, further studies documenting the range of morphological variation in each of these clades are urgently needed.If U. uncialis is considered a distinct species, then the distribution of the species, previously known from South Africa and Namibia, would include Ireland, Scotland, France, Spain, Portugal, the United States, Chile, and New Zealand based on DNA records (as Ulva sp.A, sensu Fort et al., 2021).
At least four specimens of Ulva uncialis and three specimens of U. stenophylloides have been collected from the land-based paddle raceways in the past (I&J in 2009), but none were present in any of the new collections, which suggests that U. lacinulata has likely been selected over these species under culture conditions.Alternatively, it could indicate that these species occur in low abundance or sporadically in integrated aquaculture systems.Nevertheless, it does suggest different physiological and growth potential (Fort et al., 2019) in at least two entities (U. lacinulata and U. uncialis) and further supports our decision to retain them as distinct species.
The persistence of Ulva lacinulata in culture for such a long period of time is probably due to its effective vegetative growth and its ability to withstand a range of environmental conditions, such as the contrasting temperature regimes experienced in the various sampled farms.This wide tolerance in environmental conditions is not surprising given the relatively wide distribution range of this species.At least one specimen of U. lacinulata was observed growing attached near the inlet area of the main farming complex in Hermanus, and another specimen (LK14) was observed on the seashore in Yzerfontein (west coast) in South Africa in 2009.Ulva taxonomy has been in constant flux with the application of DNA; therefore, any suggestion of introduced species in South Africa is purely speculative.It is possible that U. lacinulata might have always been present on the seashore in South Africa but was mistakenly identified as U. rigida and U. lactuca based on overlapping morphological characters with these species and the lack of DNA confirmation.Indeed, U. lacinulata was previously considered U. rigida until the type specimens of both species were sequenced (see Hughey et al., 2022).The clade also contained specimens assigned to U. lacinulata ssp.lis these two species were at one-time synonymized but were later shown to be genetically distinct (Kraft et al., 2010).The species is widely distributed and is now recorded from South Africa.Ulva stenophylloides was described from Victoria, Australia, in 2010.It was only known from Australia and New Zealand, but now also from South Africa.Ulva aragoënsis was also molecularly identified among our seashore Ulva specimens.The name U. aragoënsis was assigned to a clade containing specimens previously assigned to U. mediterranea, a replacement name for Enteromorpha aragoënsis Bliding (Alongi et al., 2014).The name U. mediterranea was however considered a superfluous name (Krupnik et al., 2018) and was subsumed into U. aragoënsis.Lastly, a foliose form of U. compressa is reported from South Africa for the first time.Studies have shown that U. mutabilis, which was synonymized with U. compressa, has a range of morphologies such as tubular, bladed and a mixed morphotype (Steinhagen, Barco, et al., 2019;Steinhagen, Weinberger, & Karez, 2019).
Our study focused on resolving the identity of commercially cultivated species of Ulva in South Africa and unexpectedly resulted in the identification of several new records and assigning a Linnean name to an unnamed clade.The most obvious next step would be to carry out a comprehensive survey of Ulva species from Southern Africa.

F
I G U R E 1 Location of commercial abalone farms and seashore sites where Ulva samples were collected.The sequence of numbers for the collection sites are arranged in ascending order from west to east.The five main Ulva-producing commercial abalone farms along the coastline of South Africa are numbered 3 (Diamond Coast Aquaculture), 6 (Abagold Abalone), 7 (Irvin & Johnson Cape Abalone), 8 (Buffeljags Abalone) and 11 (Wild Coast Abalone).The seashore sites are numbered 1 (Swakopmund), 2 (Lüderitz), 4 (Sea Point), 5 (Kommetjie), 9 (Knysna) and 10 (Eersterivier).TA B L E 1 Collection and specimen details.
J. Bolton; BMM, Brett M. Macey; MMR, Maggie M. Reddy; LK, Lineekela Kandjengo; DD, David Dyer.TA B L E 1 (Continued) F I U R E 2 Bayesian Inference tree for Ulva based on ITS rDNA region sequences (scale indicated at the bottom of the phylogeny).