Reeling them in: taxonomy of marine annelids used as bait by anglers in the Western Cape Province, South Africa

Background Common names are frequently used inconsistently for marine annelid species used as bait in the peer-reviewed literature, field guides and legislative material. The taxonomy of many such species based on morphology only also ignores cryptic divergences not yet detected. Such inconsistencies hamper effective management of marine annelids, especially as fishing for recreation and subsistence is increasing. This study investigates the scale of the problem by studying the use and names of bait marine annelids in the Western Cape Province of South Africa. Methods Fifteen recreational and six subsistence fishers at 12 popular fishing sites in the Western Cape Province donated 194 worms which they identified by common name. Worms were assigned scientific names according to a standard identification key for polychaetes from South Africa, and mitochondrial cytochrome oxidase I (COI) amplified and sequenced. Results This study identified 11 nominal species known by 10 common names, in the families Siphonosomatidae, Arenicolidae, Sabellaridae, Lumbrineridae, Eunicidae, Onuphidae and Nereididae. Cryptic diversity was investigated through employing mitochondrial COI sequences and these data will facilitate future identifications among widely distributed species. Several species (Siphonosoma dayi, Abarenicola gilchristi, Scoletoma species, Marphysa corallina, Lysidice natalensis, Heptaceras quinquedens, Perinereis latipalpa) are reported as bait for the first time, and while the names blood- and moonshineworms were consistently applied to members of Arenicolidae and Onuphidae, respectively, coralworm was applied to members of Sabellaridae and Nereididae. Analysis of COI sequences supported morphological investigations that revealed the presence of two taxonomic units each for specimens initially identified as Gunnarea gaimardi and Scoletoma tetraura according to identification keys. Similarly, sequences for Scoletoma species and Lysidice natalensis generated in this study do not match those from specimens in China and India, respectively. Further research is required to resolve the species complexes detected and also to refine the use of names by fishermen over a wider geographic range.


INTRODUCTION
In South Africa, shore-based marine fishing is an important recreational activity and part of the livelihood for many subsistence fishermen and has shown a steady increase over the last decades (McGrath et al., 1997;Sowman et al., 2014;Saayman et al., 2017). There is a close link between shore fishing and bait collecting (MacKenzie, 2005), so an increase in fishing intensity will certainly correlate with an increase in harvesting of natural stocks of bait species (Nel & Branch, 2014cf. Hodgson, Allanson & Cretchley, 2000Napier, Turpie & Clark, 2009;. However, in a recent assessment of the impacts of recreational and subsistence fishing in marine ecosystems in South Africa, impacts of bait collecting received just a passing mention (Majiedt et al., 2019). This supports Watson et al. (2017), who suggested that despite their wide use, marine annelids (i.e., polychaete worms) and probably many other bait species are universally a poorly managed resource.
A wide variety of marine invertebrates are used as bait by South African fishermen (MacKenzie, 2005;Branch et al., 2016;DAFF, 2017). However, live marine annelids (indigenous or imported) are not sold in bait shops and are instead collected by subsistence and recreational fishermen who should possess appropriate permits (DAFF, 2017). Collection is controlled by taxon-specific daily limits (DAFF, 2017), but these restrictions have remained almost unchanged for decades (cf. Gaigher, 1979;Van Herwerden et al., 1989 andDAFF, 2017). Furthermore, although nearly 2.5 million worms are harvested annually (Turpie, Heydenrych & Lamberth, 2003), biological information to inform management strategies is limited (Gaigher, 1979;Van Herwerden, 1989;Lewis, 2005;Simon et al., 2020) while restrictions on collection also do not accommodate the different bait collecting habits by recreational and subsistence fishermen . Knowing which species are being utilised is an important step towards improving management of a resource as many bait species, including those that may be morphologically very similar, may have different life history traits and habitat requirements (Hutchings & Lavesque, 2021), which may influence the vulnerability of species to exploitation. The Marine Recreational Activity Information Brochure issued by the Department of Agriculture, Forestry and Fisheries in South Africa (now the Department of Forestry, Fisheries and Environment; DAFF, 2017) identifies bait worms generically as seaworms, polychaetes and flatworms, and by various common names. The only taxa identified by genus are Arenicola Lamarck, 1801, Nereis Linnaeus, 1758, Pseudonereis Kinberg, 1865b and Gunnarea Johansson, 1927. As no images are included in the brochure, it is unclear what the worms listed by common name are. However, the popular Two Oceans: A guide to the Marine Life of southern Africa (Branch et al., 2016) provides images and common and scientific names for some baitworms: bloodworm (Arenicola loveni Kinberg, 1866), musselworm (Pseudonereis podocirra (Schmarda, 1861) as P. variegata (Grube & Kröyer in Grube, 1858)), wonderworm (Eunice aphroditois (Pallas, 1788)), Cape reef worm Western Cape Province where fishing is particularly popular (Majiedt et al., 2019), and where harvesting of worms is high (Turpie, Heydenrych & Lamberth, 2003). Furthermore, the province spans two vicariant barriers to gene flow at Cape Point and Cape Agulhas (Teske et al., 2011, Fig. 1), and this may also split species into different taxonomic units. The specific aims of the study are to: (1) identify and provide updated descriptions of the annelid species collected as bait by recreational and subsistence fishermen in the Western Cape Province of South Africa; (2) collate the common names used by the fishermen towards developing consensus for improved management; and (3) generate mtCOI sequences to explore the existence of species complexes locally and globally and facilitate identifications.

METHODOLOGY Sample sites and collection
Sampling was conducted at 12 popular beach and estuarine fishing locations in the Western Cape Province, South Africa ( Fig. 1), from June 2016 to May 2017. Collectively, these sites included sandy (Saldanha Bay, Muizenberg, Strand, Betty's Bay, Pearly Beach, Struisbaai, Witsand, Knysna) and or rocky (Velddrif, Melkbosstrand, Kommetjie, Betty's Bay, Hermanus, Witsand) habitats, which would influence the presence and absence of species collected. Worms were donated by bait collectors who all gave their prior consent to participate in the project. Involvement by most recreational fishermen was confirmed prior to sampling via fishing mailing lists or word of mouth. Some additional recreational and all subsistence fishermen were approached on site. After the aims of the study were described to participants and verbal consent received (ethical clearance number: SU-HSD-001609 from Stellenbosch University), worms were collected according to the permitted methods (DAFF, 2017), under permit RES2017-27 issued to CS by the Department of Forestry, Fisheries and Environment. Additional samples of arenicolids (bloodworm) were collected by the authors using the same techniques (see Simon et al., 2020). The common names used by the bait collectors were noted for all worms. All sampling was conducted during low tide, with specific collection methods for the different taxa included in the systematic accounts below. In some instances, fishermen were only willing to donate a small piece of the worm that was sufficient for genetic analysis.

Specimen identification and processing
Samples were relaxed in an isotonic solution of 7% MgCl 2 in tap water, measured and photographed. A section of each specimen from the mid-body or posterior was placed in 96% ethanol for molecular analysis. The rest of the specimen was fixed in 4% formalin in seawater for at least 2 days, washed in distilled water and stored in 70% ethanol. Samples were examined on Leica DM1000 light and MZ75 dissecting microscopes, and photographed with a Leica EC3 camera attachment, or on Leica DM750 light and M80 dissecting microscopes and photographed with an Olympus Targus TG5 attached to the microscope eyepieces. Where necessary, images were stacked in Helicon Focus Version 7.6.4 and processed in Photoshop Version C6. Specimens were identified using Day (1967Day ( , 1974, and where necessary, more recent literature appropriate to individual taxa. All specimens were deposited at IZIKO South African Museum (Table 1).

DNA extraction, amplification and sequencing
Approximately 25 mg of tissue was used either from mid-section or posterior end to extract DNA using the Zymo Quick DNA TM MiniPrep Plus kit (Zymo-Spin TM ) and according to the manufacturer's protocol. The universal primer pair: LCO1490 and HCO2198 (Folmer et al., 1994) was used to amplify a fragment of the cytochrome oxidase subunit 1 (COI) gene for all species. The following PCR thermal conditions were used: 94 C for 3 min; 34 cycles with 94 C for 45 s, 42 C for 1 min and 72 C for 1 min and a final extension at 72 C for 7 min (Bleidorn, Vogt & Bartolomaeus, 2005). The amplified PCR products were visualised on a 1% agarose gel using 3 ml of PCR product and 5 ml of Quick-Load Purple 100 bp DNA ladder (New England BioLabs Inc.), following Simon et al. (2020). All PCR products were sequenced using Sanger sequencing at the Central Analytical Facility at Stellenbosch University. All newly generated sequences were uploaded on GenBank (Table 1).

Molecular analysis
Sequences were edited in BioEdit Version 7.2.6 (Hall, 1999) and aligned using ClustalW with default parameters in MEGA X (Kumar et al., 2018). Neighbour joining trees were constructed in the same program, per family. Nodal support was obtained using 10 000 bootstrap replicates using the maximum composite likelihood method, with uniform rates and pairwise deletion.  Simon et al. (2020). $ The English names are listed, although fishermen frequently use Afrikaans translations: bloodworm (bloedwurm), Coral worm (koraalwurm), mussel worm (mosselwurm), moonshine worm (maanskynwurm), pudding worm (poedingwurm), wonderworm (wonderwurm). English names were never used for polwurm or bakkiewurm. Baitworm species from Western Cape, South Africa, found in this study, including common names, locations, collector details. GenBank accession numbers may be repeated when haplotypes are shared among different individuals. Samples were received from contributing fishermen and processed by Alheit du Toit (AdT), Caveshlin Naidoo (CN), Carol Simon (CS), Ethan Newman (EN) and Hendré van Rensburg (HvR). NS, no sequences.
Remarks: Specimens examined here conform to descriptions by Ashworth (1911) and Wells (1962) which included type material, but maximum size is larger. However, oval depressions seen by Ashworth (1911) ventral to some notopodia were not observed. The colour variants of A. loveni from all sites form a well-supported clade ( Fig. 7) which is exemplified by the fact that those illustrated in Figs. 4A, 4B and 4D are represented by an identical sequence (MK 922158). This clade includes two subclades, representing specimens collected on the west and south coasts, respectively. The structure seen here was previously reported in Simon et al. (2020), where nuclear data confirmed that these west and south coast clades represent a single species. The separation between these clades is demarcated by Cape Point, a location known to present a barrier to gene flow (Teske et al., 2011;Simon et al., 2020).
Remarks: Specimens examined here conform to description by Wells (1963) and Day (1967), but are smaller. Abarenicola gilchristi formed part of a distinct lineage in a well-supported clade (Fig. 7) also comprising Abarenicola brevior (Wells, 1963) and A. wellsi Darbyshire, 2017. Collection method: By hand or digging with trowel. Type locality: Buffelsbaai, Cape Peninsula, Western Cape Province, South Africa. Known distribution in South Africa: Lambert's Bay to Walker Bay. Presence in Pearly Beach extends known distribution (Day, 1967) eastwards by only a few kilometres. Namibia: Luderitz. Report in Tamil Nadu, India (Thilagavathi et al., 2013) must be treated with caution. Ecology: In sand in mid to low intertidal on sheltered shores.  Abdomen with pairs of branchiae on each segment; neuropodial lobes reduced on abdominal chaetigers, surrounded by tori; uncini with five teeth (Fig. 9J); neurochaetae verticillate chaetae (Fig. 9I); ventral cirri conical with tapering ends, becoming digitiform with rounded ends, spanning the neuropodial lobe.
Remarks: Specimens collected from all western sites (Veldrif, Betty's Bay and Hermanus) conformed to the general descriptions according to Day (1967) and Kirtley (1994). Nonetheless, differences in the morphology of the outer paleae were observed. Day (1967) described two incurving teeth present on the outer paleae, but this differs from what was observed in specimens collected in the present study: one tooth on the antero-lateral margin of the outer paleae. Additionally, Day (1967) described the inner paleae as completely concealing the "fleshy disk" or opercular disk, however, this was not observed for our specimens, instead the opercular disk was visible in the mid-section where paleae did not overlap, which was similar to Kirtley (1994). Nonetheless, all other characters observed for our specimens were similar to specimens as described by Kirtley (1994) and Day (1967), suggesting that they most likely represent Gunnarea gaimardi sensu stricto. Molecular analyses (Fig. 10) will be discussed under Gunnarea sp. 1.
Collection method: Breaking off pieces of reef by hand or narrow blade to remove worms from tubes. Type locality: Cape of Good Hope, Western Cape Province, South Africa. Known distribution in South Africa: the nominal species has been reported from KwaZulu-Natal on the east coast to the west coast of the Western Cape Province; Namibia: Walvis Bay to Luderitz (Day, 1967), but this needs to be revised. Ecology: Species forms extensive reefs by building sandy tubes on rocks in the low to mid intertidal of exposed shores. 11B). Opercular crown and opercular stalk completely fused (Fig. 11B). Two rows of golden inner and outer paleae (Fig. 11B). Approximately 34-38 inner paleae and 42-43 outer paleae. Outer paleae geniculate with a single tooth on the antero-dorsal margin (Figs. 11G2, 11G3), inner paleae geniculate with elongate, wedge-shaped peaks with a sharp tip (Fig. 11G1), and orientated toward the midline, with both rows overlapping and concealing the opercular disk (Fig. 11B). Anterior margin of opercular crown with 50 conical papillae (Fig. 11B). Pair of ciliated palps in front of mouth (Fig. 11E). Buccal lips present with upper, lower and lateral lips (Fig. 11E). Tentacular filaments compound and branched (Fig. 11E). U-shaped building organ as part of thorax (Fig. 11E); neurochaetae capillaries with bipinnate blade margins (Fig. 11C). Parathorax of three chaetigers; notochaetae alternating lanceolate and capillary chaetae (Fig. 11H), neurochaetae lanceolate chaetae of two lengths (Fig. 11I), neurochaetae thinner than notochaetae. Abdomen with a pair of branchiae on each segment; reduced neuropodial lobes surrounded by tori, uncini with seven teeth (Fig. 11F); neurochaetae verticillate (Fig. 11D). Ventral cirri conical with tapering ends, becoming digitiform with rounded ends spanning the neuriopodial lobes.
Remarks: Specimens collected from Witsand (southern site) conformed to the general description by Day (1967), including having paleae that completely conceal the "fleshy disk" or opercular disk. Specimens from the southern site generally resemble Gunnarea gaimardi (from western sites) in having a single tooth on the antero-lateral margin of the outer paleae. Nonetheless several differences were observed. Firstly, western site specimens were longer (max. of 110 mm), whereas southern specimens were a maximum of 43 mm. The most distinct feature between these two morpho-groups was the shape, orientation and arrangement of paleae on the opercular crown. The peaks of the outer and inner paleae are longer in specimens from the southern site compared to that observed in specimens from the western sites; the angle of inclination between the handle and peaks of the inner paleae is larger in western specimens than southern specimens; the outer paleae blades are wider and shorter in specimens from southern sites compared to the longer, thinner blades observed in western specimens. The inner paleae in western specimens do not overlap at the midpoint of the opercular disk, thereby exposing the disk, whereas in southern specimens the paleae overlap, completely concealing the disk. Additionally, the abdominal uncini of western specimens have five teeth, which is two less than that observed for southern specimens. Lastly, western specimens have more opercular papillae than southern specimens when comparing similar sized animals; 73, length 45 mm and 50, length 43 mm, respectively. These differences noted between specimens collected from the southern site and G. gaimardi from western sites indicate that they are indeed separate species and that specimens from the southern site (Witsand) most likely represents a new undescribed species of the genus. These morphological differences are supported by the molecular analysis which recovered two well supported clades (Fig. 10) and a genetic distance of 6% (±0.02), thus confirming their separation as independent species. The first clade, designated G. gaimardi, included specimens from Velddrif, Betty's Bay and Hermanus (western group) and the second, designated Gunnarea sp. 1, included only the specimens from Witsand (Fig. 10). Morphological differences together with the genetic separation of the clades indicate the presence of two species in what has, till now, been considered a monospecific genus (Capa, Hutchings & Peart, 2012). Preliminary observations of Gunnarea sp. from Port Shepstone in KwaZulu-Natal suggest that they conform to the description of Gunnarea sp. 1 and studies are underway to confirm this. Description: more than 145 mm; L10 = 8.1 mm, W10 = 3.4 mm (Fig. 12A). Prostomium conical, peristomium with two rings, second slightly shorter than first (Fig. 12A) (Fig. 12H), appearing white. Aciculae yellow. Dental formula: MI = 1 + 1, MII = 5 + 5, MIII = 2 + 2, MIV = 1 + 1 (Fig. 12B), MV free, lateral to MIV and MIII.

Remarks:
The specimen conforms to the general description of S. tetraura according to Schmarda (1861) and Day (1967), and no characters could be identified to distinguish the specimen collected here from the description of specimens from Chile. However, the specimen differs morphologically and genetically from others collected in this study that also conform to the description of S. tetraura by Day (1967), and genetically from specimens identified as S. tetraura in China ( Fig. 13; discussed under Scoletoma sp. 2 below). The species is therefore identified here as Scoletoma sp. 1.
Remarks: All six specimens conform to the general description of S. tetraura according to Schmarda (1861) and Day (1967), and no characters could be identified to distinguish the specimens collected here from the description of specimens from Chile. However, this species differs from the specimen from Betty's Bay. In Scoletoma sp. 2 from Hermanus, the long-headed simple hooded hooks are about 25% shorter than those of Scoletoma sp. 1 from Betty's Bay, and post-chaetal lobes are about 30% longer in the posterior. Furthermore, the segments of Scoletoma sp. 2 appear to be longer than those of Scoletoma sp. 1; in specimens that are similarly wide, specimens of the former are 1.5 to 1.8 mm longer for the first 10 chaetigers than in the latter. Finally, specimens of the two species were collected from different habitats. Further research is needed to determine which, if any, refers to the species recorded previously by Day (1967) as S. tetraura. The morphological separation is supported by molecular analyses (Fig. 13) that retrieved two well-supported operational taxonomic units, Scoletoma sp. 1 (from Betty's Bay) and Scoletoma sp. 2 (from Hermanus). The two Scoletoma species from South Africa form part of a weakly supported clade together with Scoletoma fragilis (O.F. Müller, 1776), Lumbrineris aberrans Day, 1963, Lumbrineris erecta Moore, 1904, Lumbrineris japonica Marenzeller, 1879, and Lumbrineris perkinsi Carrera-Parra, 2001 which is separate from S. tetraura from China. The separation of Scoletoma spp 1 and 2 from South Africa and S. tetraura from China in two different clades with high support suggests that they are independent species. However, without sequences from the species' type locality in Chile, it is impossible to determine whether the specimens found in China and South Africa all represent new species or whether one of them is an alien. Specimens from the extended global distribution of S. tetraura need to be examined, as there are likely more species within this complex. Additionally, S. tetraura and S. fragilis were previously considered members of Lumbrineris, so the other Lumbrineris species in the clade should be revised to determine whether they are also in the genus Scoletoma, or whether this genus is paraphyletic.
Collection method: Samples from Hermanus collected among broken pieces of Gunnarea tubes.
Remarks: Specimens collected in this study conform to the general description according to Day (1967). Unfortunately, the original description of M. corallina (Kinberg, 1865a) was poor, with no illustrations against which to compare the specimens collected in this study. However, since the type locality of M. corallina is in Hawaii and the species has a global disjunct distribution, it is probable that the specimens collected here are really an incorrectly identified indigenous species. We therefore take the more conservative route and refer to the species collected in South Africa as M. cf. corallina. All specimens collected during this study were incomplete, missing their posterior ends, so characters such as anodont chaetae, the number of branchial filaments and the number of aciculae in the posterior regions were not documented and thus could not be commented on. All sequences generated clustered with M. corallina from KwaZulu-Natal (KT823410) (Kara, 2015), with high bootstrap support, indicating that it is a single species (Fig. 16). Further investigation is underway to confirm the taxonomic status of M. corallina in South Africa.
Collection method: By hand from sediment under rocks. Known distribution in South Africa: Mabibi in northern KwaZulu-Natal to Mgazana in the Eastern Cape Province, Witsand in Western Cape Province (Day, 1967;current study). Apparent distribution globally: Mozambique, New Zealand, Red Sea, Australia, Marshall Islands, Lakshadweep Island and Juluit Atoll (Day, 1967;Read & Fauchald, 2021). Ecology: Occupies burrows in sediment under rocks in the mid-intertidal zone.
Remarks: Specimens collected here conformed to the description by Kara et al. (2020), except for those collected from Knysna and Betty's Bay which have a blue anterior (approximately first six chaetigers), becoming light brown in the middle to posterior end. Phylogenetic analysis recovered a single well-supported clade that comprised all specimens from Knysna, Betty's Bay, Strand, Kommetjie (Soetwater) and Melkbosstrand, indicating that the colour morphs are a single species (Fig. 16). Lewis & Karageorgopoulos (2008) observed colour variation in specimens which included iridescent blues and greens for the reproductive segments along the length of the body, from chaetigers 70-80. This does not conform to the colour morphs found in the specimens in the present study in which the colour was observed in the anterior regions. Nonetheless, the colour on the remaining parts of the body, "medium brown in the middle and darker toward the posterior", conform to that reported by Lewis & Karageorgopoulos (2008). The use of two species of Marphysa in the Western Cape Province supports recent research showing that globally, multiple species of this genus, especially members of the M. sanguinea complex, are used as bait, even within regions (see review by Hutchings & Lavesque, 2021). Although the current study showed that different colour morphs represent a single species, further research is needed to determine whether individuals occupying different habitats, as described by Day (1967) and Lewis & Karageorgopoulos (2008), are also a single species.
Collection method: By hand from sediment under boulders in boulder fields. Type locality: Cape of Good Hope, Western Cape, South Africa. Known distribution in South Africa: Langebaan Lagoon on the west coast to Port Elizabeth on the south coast (Day, 1967;Kara et al., 2020). Ecology: Occupies burrows in sediment typically grey/black medium to coarse grains and rich in sulphur. In Knysna, specimens were found in sandier sediments.  Lysidice natalensis Kinberg, 1865a: 566;Hartman, 1948: 84, 85, Pl. XI Figs. 1-2;Day, 1951: 40;Day, 1953: 435;Day, 1960: p 336;Day, 1967: 401, Fig. 17.7 K-R;Branch et al., 2016: 70, Fig. 26.9 Lysidice atra Schmarda, 1861 Lysidice capensis Grube, 1868: 12, Fig. 4 Description: Complete specimens 62 and 63 mm long for 126 and 156 chaetigers. L10 5.28-9.8 mm, W10 1.84-4.5 mm. Colour reddish-brown with white spots, both extending into middle of prostomium and antennae, margin of prostomium and tips and base of antennae white (Fig. 18A). Prostomium bilobed, antennae tapered, lateral antennae shorter than prostomium, median antenna slightly longer, proximal part brown, tips white (Fig. 18A). Mandibles thick; MI 1 + 1; MII 3 + 3; MIII 2-3 + 0; MIV 2-3 +4-7; MV 1 + 1. Parapodia with slender dorsal cirri (Fig. 18B), becoming shorter and thinner from chaetiger 22 to 38 onwards (Figs. 18E, 18H). Ventral cirrus bluntly triangular (Fig. 18B), getting shorter posteriorly (Fig. 18E), nipple-shaped in posteriormost chaetigers (Fig. 18H). Post-chaetal lobe truncate (Fig. 18B), getting shorter posteriorly (Fig. 18E), inconspicuous in posteriormost chaetigers (Fig. 18H). Superior chaetae limbate capillaries and comb chaetae of two sizes (Fig. 18F). Inferior compound chaetae with short blades, bidentate, teeth usually of similar sizes (Figs. 18C, 18D), but proximal tooth may be thicker and or longer. Acicula black with blunt tips, one in anterior chaetigers, two in middle and posterior (Figs. 18B, 18E, 18H); bidentate acicula hook with small hood from chaetiger 25-28 onwards (Figs. 18E, 18H), teeth may be worn, giving unidentate appearance (Fig. 18G). Kinberg (1865a) is poor, but this material is later described by Hartman (1948). Specimens collected here generally match this latter description, and those by Day (1951Day ( , 1953Day ( , 1967, although the posterior ventral cirrus is more prominent than described by Day (1967). The wide distribution within South Africa is suggestive of multiple species and may be further reflected by the two species that Day (1967) synonymised with L. natalensis without explanation. It is therefore possible that L. capensis and L. atra, both originally described from the temperate Western Cape Province in Kalk Bay and the Cape of Good Hope, respectively, are not L. natalensis which was first described from Durban in the subtropical KwaZulu-Natal. Additionally, Day (1967) provides no explanation for why L. atra, which was described four years before L. natalensis and therefore claims priority, was synonymised with the latter. More specimens from throughout the distribution range and any available type material need to be examined to resolve the taxonomy of this species. The description of L. natalensis from Pakistan by Mustaquim (2000) is not very detailed, and the only differences from samples examined here are differently shaped post-chaetal lobes. All specimens from Witsand form a well-supported clade that is not reciprocally monophyletic with L. natalensis from India ( Fig. 16; Sigamani et al., 2020). Identity of the species in Pakistan is also doubtful.

Remarks: Original description by
Collection method: By hand. Type locality: Durban, KwaZulu-Natal, South Africa. Known distribution in South Africa: From Namibia to northern KwaZulu-Natal (Day, 1967). Ecology: Habitat variable; in the current study specimens were collected from under rocks in rock pools, Day (1934) reported them from muddy sand.
Prostomium with frontal extension forming palpohores for frontal palps (Fig. 19C). Lateral antennae reaching chaetiger 4-7 on posterior part of prostomium, shorter median antenna reaching chaetiger 2-4 placed anterior to lateral antennae. Proximal ceratophoral rings wide, covering most of prostomium (Fig. 19A). Ceratophores with 15-30 rings on median antennae and 20-48 rings on lateral antennae, each terminating in an elongated distal ring. Ceratophores at least as long as styles but up to twice the length of styles which taper distally (Figs. 19B,19D). Peristomium as long as, or longer than, prostomium with Figure 19 Morphology of Heptaceras quinquedens (Day, 1951). (A) Dorsal anterior of preserved specimen showing peristomial notch flanking prostomial ridge, laterally curving peristomial cirri and iridescent shine that remains after preservation, (B) Dorsal anterior of live specimen, insert shows freckled spots on anterior dorsum, (C) Ventral anterior of preserved specimen, (D) Lateral anterior view of preserved specimen showing cylindrical shape of modified parapodia and progression of deep mid-dorsal notch on the dorsal margin, flanking an elevated prostomial ridge (Fig. 19A). Peristomial cirri as long as peristomium, slender and tapering, situated distally on peristomium on either side of the mid-dorsal notch, curving laterally (Figs. 19A,19B).
Remarks: The specimens examined here match earlier descriptions (Day, 1951(Day, , 1967Fauchald, 1982), but this is the first observation of tridentate falcigers in the modified parapodia, although tridentate falcigers are known to occur within the genus (Paxton, 1986). The third tooth is small (Fig. 19G) and not always present so can easily be overlooked. According to Fauchald (1982) the median antenna is longer than the posterior lateral ones in the holotype (reaching chaetiger three vs. two) but in all of the material examined here, the posterior lateral antennae were longer than the median antenna, conforming to the description by Paxton (1986). The iridescent shine seen on the body of H. quinquedens is similar to that of Diopatra aciculata (Van Rensburg, Matthee & Simon, 2020) and may be why fishermen commonly refer to both species as moonshineworms.
Remarks: Specimens collected in this study conformed to the recent redescription in Villalobos-Guerrero (2019). However, variation in body size and paragnath arrangement was noted; total length of paratype is 127 mm and paragnath arrangement, Area III = 9, Area IV = 18-23, Area VII-VIII = 53.
Collection method: From under rocks in the mid-intertidal zone. Type locality:   Description: Body length up to more than 140 mm. Colour variable: greenish-brown, greyish-brown and medium brown (Figs. 21A, 21B) with white pigmented spots around 4 eyes on prostomium. Black pigmented spots along midpoint of segment boundaries from chaetiger 13 (Fig. 21B). A mix of different types of paragnaths; conical, shield-shaped and p-bars; arranged in distinct areas on pharynx. Area I = 1 conical, Area II = 15-17 conical in a wedge shape, Area III = 22 conical in three or four rows, Area IV = 27-32 conical and p-bars in a closely spaced arc shape, Area V = 1 conical, Area VI = large shield-shaped bars and Area VII-VIII = 40 conical and p-bars alternating in 2-4 rows (Fig. 21C). Oral ring (Fig. 21C), AVI-V-AVI pattern, y-shaped: ridges of AVI sub-medially separated producing parallel furrows. Notopodial ligule enlarged and elongated from chaetiger 13 to posterior (Fig. 21F). Dorsal and ventral cirri present (Fig. 21F). Homogomph spinigers with finely serrated blades (Fig. 21E) and heterogomph falcigers ( Fig. 21D) with concaved and finely serrated blades.
Remarks: Specimens collected in the study conformed to the redescription in Kara, Macdonald & Simon (2018), except for body length which was larger, measuring up to a maximum of 140 mm. Molecular analyses (Fig. 22) recovered a single monophyletic group with strong maximum likelihood support, indicating a single genetically similar population, further supporting Kara, Macdonald & Simon (2018). Synonymy of P. podocirra with P. variegata was recently reversed (Kara, Macdonald & Simon, 2018), but it is not known whether P. variegata in KwaZulu-Natal in South Africa, Namibia and Mozambique, as reported by Day (1967), are a single species.
Collection method: Breaking off mussels by hand from the mussel bed, or by pouring household bleach over the bed (A. du Toit, 2017, personal observation). Collection of nereidid species is no longer permitted (DAFF, 2017). Type locality: Cape of Good Hope, Western Cape, South Africa. Known distribution in South Africa: Lamberts Bay to Kidds Beach (Kara, Macdonald & Simon, 2018), possibly extending up the east coast to KwaZulu-Natal and Mozambique and up the west coast to Namibia (Day, 1967). Ecology: In low intertidal among mussel beds and abandoned Gunnarea tubes and barnacle shells.

DISCUSSION
This study found that more marine annelid taxa are utilised in South Africa as bait than what has previously been reported. In addition to the widely reported and investigated bait species (Arenicola loveni, Gunnarea gaimardi, Marphysa haemasoma and Pseudonereis podocirra; e.g., Van Herwerden, 1989;Lewis, 2005;Sowman, 2006;Lewis & Karageorgopoulos, 2008;Branch et al., 2016), several taxa were recorded for the first time (Abarenicola gilchristi, Gunnarea sp. 1, Heptaceras quinquedens, Lysidice natalensis, Marphysa cf. corallina, Perinereis latipalpa, Scoletoma spp 1 and 2). This is also the first published report of Siphonosoma dayi being used, even though there have been anecdotal reports of fishermen collecting sandworm in Knysna since at least 2009 (M.K.S. Smith, 2021, personal communication). By contrast, Arabella iricolor and Eunice aphroditois (or species matching their general descriptions), which are listed as bait in legislation and field guides (Marine Living Resources Act, 2014; Branch et al., 2016), were not collected in this study. This suggests that more species are used in the province than collected by us, possibly because these species did not occur at the sites sampled, and or that identifications of these species being used were incorrect. For example, one of the authors never found E. aphroditois in the Western Cape Province even after extensive sampling in apparently appropriate substrate, although she did find Eunice species in the subtidal in KwaZulu-Natal on the east coast (J. Kara, 2019. Furthermore, A. iricolor is superficially similar to lumbrinerid species, and it is possible that both taxa are collected, or that these species were confused in the records for bait collecting. Finally, it is not possible to determine whether species that were collected at single sites (L. natalensis and P. latipalpa) are targeted more widely, or were misidentified since both were called by names more widely used for other species. Including Diopatra aciculata collected in Knysna in a parallel study (Van Rensburg, Matthee & Simon, 2020), 14 species were identified in the Western Cape Province by ten common names, excluding Afrikaans translations. For species collected multiple times and from different locations, individual common names were sometimes applied to more than one species. Species of the same family or genus were often known by a single common name; for example, arenicolids (Arenicola loveni and A. gilchristi) are bloodworm, onuphids (D. aciculata and H. quinquedens) are moonshineworms, Scoletoma species are puddingworms and Marphysa species are wonderworms. For the arenicolids and onuphids this is true even when the species show clear morphological or environmental differences which may have been noted by fishermen, as evidenced by fishermen in Pearly Beach who distinguished between bloodworm (A. loveni) and the bakkiewurm (A. gilchristi). This was the first time that a second arenicolid is reported as bait, even though DAFF (2017) acknowledges that more than one species may be used when they specify that bloodworm are "All species of the genus Arenicola", although this is inaccurate as only one species of Arenicola has been recorded locally. Individual species were sometimes called by multiple common names that were not translations of the same thing. For example, M. haemasoma was identified as wonderworm, bloodworm or bloukoppie (this is Afrikaans for 'blue head', referring to the blue anterior of worms from Knysna and Betty's Bay); G. gaimardi was identified as coralworm and polwurm ('pol' is Afrikaans for a tuft, tussock or clump of grass, and may here refer to the clumps of tubes formed by the worms); P. podocirra was identified as musselworm and coralworm, while D. aciculata was also called the pypiewurm (this is Afrikaans for 'pipe worm', undoubtedly alluding to the chimneys that extend from the mouths of the tubes) by bait collectors in Port Elizabeth (H. van Rensburg, 2017, personal observations). It is also apparent that individual common names were sometimes applied to species from different families, such as coralworm (G. gaimardi, P. latiplapa¸P. podocirra) and musselworm (P. podocirra, L. natalensis).
For the most part, subsistence and recreational fishermen used the same names (e.g., for arenicolids, sabellarids, onuphids and Marphysa species). Variations in use of names may suggest unfamiliarity with bait worms among some subsistence fishermen, such as bloodworm for M. haemasoma in Melkbosstrand, or differences in the use of names depending on geographic region and or type of fishermen, such as coralworm for nereidids at Kommetjie and Velddrif. Interestingly, none of the fishermen used the names from Branch et al. (2016) for M. haemasoma (estuarine wonderworm which distinguishes it from E. aphroditois, the wonderworm), G. gaimardi (Cape reef-worm), L. natalensis (three-antennaed worm) or Scoletoma species (S. tetraura false earthworm). Finally, several common names that appear in DAFF (2017), such as rock, shingle, or pot worms, were not used for any of the species collected in this study. The results of this study confirm that common names are sometimes applied in an inconsistent manner by managers and bait collectors. These differences may be maintained through the transfer of knowledge, across generations of bait collectors, of the identification of worms by morphology and ecological patterns. However, it is possible that the application of common names has changed (e.g., the name moonshineworm applied to onuphids and not A. iricola (Marine Living Resources Act, 2014)).
The genetic data confirmed the presence of complexes of morphologically similar species within South Africa and globally. Day (1967) reported Gunnarea gaimardi and S. tetraura from Namibia to northern KwaZulu-Natal. Given that this range spans the cold Namaqua, warm Agulhas, and subtropical Natal ecoregions (Sink et al., 2012) and barriers to gene flow at Cape Point, Cape Agulhas, Algoa Bay and Wild Coast (Teske et al., 2011), it is not surprising that these nominal species each included two genetically distinct species with geographic and habitat separation, respectively. This may also apply to L. natalensis that has a similar distribution (Day, 1967;Branch et al., 2016). Even though all specimens identified here as Gunnarea (including Gunnarea sp. and G. gaimardi) and the Scoletoma species from Hermanus and Betty's Bay matched the descriptions of the nominal species provided in Day (1967; gaimardi and S. tetraura, respectively), the two genetic groups identified in each could be easily distinguished after thorough morphological examination. This supports Hutchings & Kupriyanova (2018) who suggested that many descriptions contained in Day (1967), especially of species described before the 1900s such as the two species under discussion, are too generic to enable accurate identification. Similarly, sequences of L. natalensis and Scoletoma species 1 and 2 generated in this study do not match those generated for L. natalensis and S. tetraura collected in India and China, respectively (Zhou et al., 2010;Sigamani et al., 2020), indicating the presence of complexes of species that may be morphologically similar but genetically distinct, from different locations around the world. Sigamani et al. (2020) used Day (1967 to identify their samples which also included H. quinquedens, originally described from South Africa; unfortunately, we were unable to obtain sequences for the samples that we gathered to test whether the specimens from the two countries are conspecific. However, our results again support Hutchings & Kupriyanova (2018) who warned that using Day (1967) to identify polychaetes outside of southern Africa may erroneously inflate the distribution ranges of polychaete species.
Resolving the identities of marine annelids used as bait has several important management implications. This is exemplified by the recent discovery that moonshineworm collected in Swartkops and Knysna estuaries is D. aciculata, a species originally described in Australia and is probably an alien in South Africa (Elgetany et al., 2020;Van Rensburg, Matthee & Simon, 2020). Thus, the focus of management of this species must change from conserving populations to preventing further population growth and spread (Van Rensburg, 2019;Van Rensburg, Matthee & Simon, 2020). This could be done by permitting increased removal by bait collectors, but preliminary investigations suggest that this is unfeasible (Van Rensburg, 2019) and that alternative management strategies need to be explored. Knowing the identity of the worms used may also have important implications for the movement of bait species between sites where worms are collected and where fish are caught, since unused bait that can regenerate is frequently discarded in the latter (M.K.S. Smith, South African National Parks, Knysna). This is especially important if the species is alien (as D. aciculata), or if species thought to be locally widespread are multiple species with restricted distributions (as may be the case for Gunnarea, Scoletoma and Lysidice species).
The disjunction between the common names used by collectors and managers is especially problematic when considering the worms that should not be collected. The most recent brochure issued by DAFF (2017) states that Cape reef worm (specified as Gunnarea), cannot be collected, but that coralworm can. Since collectors contributing to this study all called Gunnarea coralworm, and because it is unlikely that many would know the genus name, bait collectors could collect this species not knowing that they are breaking the law (or use it as a defence if they do). The prohibition on collection of Gunnarea and musselworms (identified as Nereis and Pseudonereis by DAFF (2017)) is related to the structural damage caused to reefs and mussel beds during collection (Van Herwerden, 1989), although this is not clearly articulated in the information brochure. It may therefore be more effective to specify the prohibition of taxa based on the habitats that they occupy, and not just name.
This study was limited by several constraints. Firstly, the geographical coverage was restricted relative to the total coastline of the province; the fishing sites were selected according to where participants could be recruited in advance (because bait collecting is time consuming and needs to coincide with low tides which further limited sampling opportunities, we contacted a core of the participants via fishing mailing lists to ensure success in collection) while we also avoided sites that were potentially unsafe, such as Strandfontein and Monwabisi beaches along the northern shores of False Bay. Because of this sampling strategy, there was a bias towards recreational fishermen because subsistence fishermen could not be contacted in advance. Instead, subsistence fishermen were approached on an ad hoc basis if they were active at the preselected sampling sites. Additionally, many subsistence fishermen were unwilling to donate bait to the project because bait collecting is so time consuming. We were also reluctant to offer compensation to fishermen because the sale of worms is prohibited by law (Marine Living Resources Act, 2014). Consequently, our understanding of the use of common names is still incomplete because species reportedly used as bait, but not found, could not be addressed in this study. This is further exacerbated when fishermen from different fishing sectors and who speak different languages (e.g., English, Afrikaans, isiXhosa) use different names.
In conclusion, the current study has confirmed that more polychaete species are currently used as bait than previously reported. Furthermore, the inconsistent application of common names across taxa and among users, including for the more popular and widespread species, may hamper effective management. The detection of pseudocryptic species complexes among some bait species may have further implications for the management of these taxa as individual species should form separate management units, especially if they are spatially separated. Finally, diversity of marine annelids in general, and bait species in particular, has been underestimated in South Africa, and the global distribution of some has been overestimated. Research to clarify the taxonomy of the members of the pseudocryptic species complexes identified here, i.e., Scoletoma species 1 and 2 and Gunnarea sp. 1, and the use of polychaetes and common names across a wider geographic range is ongoing.