Triakidae, J.E.Gray, 1851

Elasmobranch, Its Implications For Global, Parasitology, Diversity And, Naylor, G. J. P., Sc, Caira, J. N., Ct, Jensen, K., Ks, Rosana, K. A. M., Fl, White, W. T., Csiro, Tas, Last, P. R., Csiro & Tas, 2012, A Dna Sequence-Based Approach To The Identification Of Shark And Ray Species And Its Implications For Global Elasmobranch Diversity And Parasitology, Bulletin of the American Museum of Natural History 2012 (367), pp. 1-262 : 35-41

Triakidae (houndsharks)

Mustelus (smoothhounds) ( figs. 23, 24)

This was among the most problematic of genera included in our analysis. At present, the genus includes,31 species of which five have been described since 2005; clearly a substantial amount of diversity in this genus remains to be explored. Analysis of 129 specimens yielded 18 clusters, each of which we believe represents a distinct species of Mustelus . However, assigning correct names to clusters without vouchers was problematic; particularly challenging were the specimens collected from the Gulf of California. For nine clusters, photo vouchers are available for one or more of the included specimens. For three additional clusters, identifications were based on one or more museum specimens. In the remaining six cases we have relied on geographical range and/or the taxonomic expertise of individuals providing samples. Ultimately, while we are confident in the existence of 18 species among these specimens, the specific identifications of many of the clusters remain to be confirmed. We hope that the generation of comparable sequence data for additional specimens of known identity will serve to place the data presented here into a broader perspective. Each of the 18 species clusters is treated separately below.

Mustelus widodoi (whitefin smoothhound) and

Mustelus ravidus (Australian gray smooth-

hound) ( fig. 23)

In total, 31 specimens were analyzed, consisting of 30 specimens of M. widodoi from Borneo and one specimen of M. ravidus from a specimen deposited in the Australian National Fish Collection (GN4898 5 ANFC H 5947-01). The analysis yielded a cluster consisting of the 30 M. widodoi specimens from Borneo, with the M. ravidus specimen from Australia grouping outside. The specimens from Borneo were consistent morphologically with M. widodoi of White and Last (2006); however, they were collected further north than the type specimens and represent a distributional extension for this species (see Last et al., 2010c). The range of pairwise differences among the M. widodoi specimens was 0–5; the average was 1. The average of the pairwise differences between M. widodoi and M. ravidus was 8.5. We note that M. ravidus was referred to as Mustelus sp. A by Last and Stevens (1994) and Gardner and Ward (2002).

Mustelus mosis (Arabian smoothhound) ( fig. 23)

The three specimens included in the analysis were collected from India and thus represent the eastern region of the distribution of this species, which extends from India, westward to Somalia, and possibly as far southwest as South Africa. The analysis yielded a single cluster. The range in pairwise differences within this cluster was 3–9; the average was 6.7. This is one of only two species of Mustelus known to occur in India. These specimens were found to group well outside those of the other Indian species, M. manazo , the identity of which is grounded by specimens.

Mustelus mustelus (smoothhound) ( fig. 23)

In total, nine specimens of this species were included in our analysis, three from Senegal, one from Angola, and five from South Africa. Thus, these specimens come from much of the known distribution of this species, which extends from the North Sea along the coast of Europe and the western coast of Africa to South Africa. The analysis yielded a single cluster, with some evidence of substructure. The range in pairwise differences among all nine specimens was 0–8 ; the average was 3.4. The range in pairwise differences among the specimens from Senegal was 1–3 (with an average of 2). The range among the other five specimens was 0–1 (with an average of 0.3). The average of the pairwise differences between specimens from Senegal and those from southern Africa was 6.2. The identities of a number of specimens in this cluster are grounded with photographs .

Mustelus cf. lunulatus (sicklefin smoothhound)

( fig. 23)

The analysis yielded a cluster of eight specimens that we have tentatively identified as M. cf. lunulatus . The range in pairwise differences among specimens was 0–7, with an average of 2.9. The identity of this cluster is problematic. All eight specimens were collected from the Gulf of California and all possess teeth with low, rounded cusps. They were initially identified as M. lunulatus , but dissections and x-rays revealed a number of inconsistencies. For example, while two of the specimens (BJ-802 and BJ-796) have precaudal vertebral counts (77 and 81, respectively) that are within the range for this species (74–82) presented by Heemstra (1973), their monospondylic vertebral count (37 in both cases) is outside the range for M. lunulatus (28–34) presented by Heemstra (1973). Moreover, the cluster includes a specimen (BJ-803) that has a precaudal vertebral count (87) that falls within the range for M. intermedius (83–93) of Heemstra (1973) but has a monospondylous vertebral count (41) that falls within the range (41–44) for Heemstra’s ‘‘ M. platyrhinus . ’’ The identity of this cluster should be considered tentative until confirmed specimens of M. lunulatus can be included in the analysis.

Mustelus canis (dusky smoothhound) ( fig. 23)

The analysis included eight specimens identified as this species; seven were collected from the western North Atlantic coast from Rhode Island to North Carolina and one from the Bahamas. Only some of the Northern Hemisphere elements of the distribution of this species are represented and the Southern Hemisphere elements are unrepresented. The analysis yielded a single cluster, with the specimen from the Bahamas grouping outside those from the more northern latitudes. The range in pairwise differences among the eight specimens within this cluster was 0–23; the average was 7.9. Some consideration should be given to the possibility that additional species of Mustelus may be represented by these specimens. Heemstra’s (1997) recognition of Mustelus canis insularis from several islands of the Caribbe- an, may, for example be relevant to the identity of the specimen included here from the Bahamas.

Mustelus sp. 2 ( fig. 23)

The analysis yielded a cluster of four specimens from the Gulf of California with a range in pairwise differences among specimens of 0–5 and an average of 2.5. We have been unable to definitively identify this cluster to species. Like candidate species from the Gulf of California, these specimens possess teeth with low rounded cusps. However they lack the white posterior margins of the dorsal, pectoral, pelvic, and anal fins described by Castro-Aguirre et al. (2005) in M. albipinnis and also described by Pérez-Jimé nez et al. (2005) in M. hacat , a potential junior synonym of the former species (Eschmeyer and Fricke, 2011). Moreover, all four specimens were collected by artisanal fishermen using bottom gill nets and thus, unlike specimens of M. albipinnis , came from waters substantially shallower than 100 m. While one of the specimens (GN1565 5 BJ- 671) in this cluster has a monospondylic vertebral count (41) that is generally consistent with that of M. albipinnis (40), it has a precaudal vertebral count (97) that is above that reported by Castro-Aguirre et al. (2005) for M. albipinnis (i.e., 92), and slightly lower than the range (101–102) reported in the original description of M. hacat . The possibility that this species represents one of the two undescribed species from the Gulf of California treated by Heemstra (1973) should not be ruled out.

Mustelus henlei (brown smoothhound) ( fig. 23)

Material from the Gulf of California also included six specimens of Mustelus henlei . These specimens represent only a small portion of the distribution of this relatively easy to identify species, which has been reported from as far north as Washington state and as far south as Peru. Analysis of these specimens yielded a single cluster with a range in pairwise differences of 2–9, and an average of 5.6. The identity of this cluster is grounded with images.

Mustelus norrisi (narrowfin smoothhound) ( fig. 23)

A single specimen, collected and identified by NOAA Fisheries biologist Lisa Jones, was included in the analysis. This specimen grouped independently of all of the other specimens of Mustelus included in the analysis. However, clearly this specimen represents only one point in the distribution of this species which extends along the eastern coast of North and South America, from Florida to Venezuela, and Argentina.

Mustelus californicus (gray smoothhound) ( fig. 23)

Our analysis included 15 specimens of this species, all collected from the Gulf of California. Given that this species occurs along coastal California, the Gulf of California, and possibly as far south as Puerto Vallarta, our specimens represent the center of the relatively limited overall distribution of this species. Three of these specimens were deposited in the Texas Cooperative Wildlife Collection (GN2284 5 TCWC 7561.03, GN5291 5 TCWC 7561.05, and GN5292 5 TCWC 7561.04). The analysis yielded a single tight cluster. The range in pairwise differences among specimens within this cluster was 0–2 and the average of the pairwise differences among specimens was 1.4. The identity of this cluster is grounded not only with photographs, but also in that dissections and/or x-rays revealed that specimens had an extra symphyseal cartilage on the palatoquadrate, which other sympatric Mustelus species lack ( Compagno, 1984a).

Mustelus sp. 1 ( fig. 23)

The analysis yielded an additional cluster of two specimens from the Gulf of California, which were identical in sequence. We have been unable to definitively identify this cluster to species. These specimens, like those of Mustelus sp. 2 , possess teeth with low, rounded cusps and lack the white posterior margins of the dorsal, pectoral, pelvic and anal fins described by Castro-Aguirre et al. (2005) in M. albipinnis as well as in its potential junior synonym M. hacat (see Pérez-Jiménez et al., 2005). Like the four specimens of Mustelus sp. 2 , both specimens were also collected by artisanal fishermen using bottom gill nets and thus, unlike M. albipinnis , came from waters substantially shallower than 100 m. The possibility that these specimens represent one of the two undescribed species from the Gulf of California treated by Heemstra (1973) can also not be ruled out. These specimens clustered together, independent of the remaining Mustelus species. They clustered most closely with M. californicus ; the average of pairwise differences between these two species was 40.1.

Mustelus antarcticus (gummy shark) ( fig. 24)

A total of nine specimens of this species were included in the analysis. All nine came from southeastern and southwestern Australia and are representative of much of the range of this species. The analysis yielded a single cluster, with a range in pairwise differences of 0–4 and an average of 1.9. The identity of this cluster is based on a specimen in the Australian National Fish Collection (GN4902 5 ANFC H 6571-03).

Mustelus lenticulatus (spotted estuary smooth-

hound) ( fig. 24)

This species occurs throughout New Zealand, which was the site of collection of all three specimens included in the analysis. The range in pairwise differences seen within the cluster that resulted from the analysis was 0–1. The identity of this cluster is based on two specimens in the Australian National Fish Collection (GN4896 5 ANFC H 5551- 01 and GN4897 5 ANFC H 5551-02).

Mustelus stevensi (whitespotted gummy shark)

( fig. 24)

Two specimens of this recently described species (see White and Last, 2006) were included in the analysis. Both specimens were collected from the type locality, from the Dampier Archipelago off northwestern Australia. These clustered together, independently of the remaining Mustelus species. The sequences of the two specimens differed by 2. The identity of this cluster is confirmed given that one of the samples comes from a paratype (GN4900 5 ANFC H 4650-01); the second sample comes from a voucher (GN4899 5 ANFC H 4649-08). The average of the pairwise differences between M. stevensi and M. lenticulatus was 14.3, between M. stevensi and M. antarcticu s was 17.2, and between M. stevensi and M. ravidus was 83. Our results support the clear distinctions reported by Gardner and Ward (2002) between M. stevensi (5 Mustelus sp. B of Last and Stevens, 1994, and Gardner and Ward, 2002) and each of M. lenticulatus , M. antarcticus , and M. ravidus (5 Mustelus sp. A of Last and Stevens, 1994, and Gardner and Ward, 2002).

Mustelus manazo (starspotted smoothhound)

( fig. 24)

In total, 13 specimens of this species were included in the analysis. These consist of seven specimens from Taiwan, two from Vietnam, three from Japan, and one specimen that was landed in Borneo. The analysis yielded a single cluster. The range in pairwise differences within this cluster is 0–6 and the average was 2.8. However, our specimens represent only the eastern central elements of the distribution of this species. Four samples from this cluster are represented by specimens (GN972, GN989, GN1019 and GN1020 5 UMMZ 231357).

Mustelus palumbes (whitespot smoothhound)

( fig. 24)

Six specimens of this species, all collected from South Africa, were included. These are representative of the distribution of this species which occurs from Namibia to central Natal. The range in pairwise differences of this cluster was 0–4, and the average was 2.6.

Mustelus asterias (starry smoothhound) ( fig. 24)

The analysis included two specimens, both collected from the eastern North Atlantic. The sequences of these two specimens differed by one base. However, our specimens represent only the northernmost region of the distribution of this species, which extends along the coast as far south as Senegal. These specimens clustered most closely with those of M. palumbes . The average of the pairwise differences between specimen of these two species was 8.3.

Mustelus schmitti (narrownose smoothhound)

( fig. 24)

In total, seven specimens of this species, all collected and identified by Gustavo Chiaramonte of the Museo Argentino de Ciencias Naturales, Buenos Aires, were included. The analysis yielded a single cluster, with a range in pairwise differences among specimens of this species of 0–4 and an average of 1.9. These specimens are fairly representative of the distribution of this species, which extends along the coasts of southern Brazil, Uruguay, and Argentina.

Scylliogaleus quecketti (flapnose houndshark)

( fig. 24)

The analysis included four specimens of this species, all from South Africa ; one donated by Mark Harris and three others by Geremy Cliff of the KwaZulu-Natal Sharks Board. The analysis resulted in a single cluster of all four specimens of this monotypic genus. These specimens are representative of the distribution of this species, which is known only from the southeastern coast of South Africa. The range in pairwise differences among specimens within this cluster was 0–2 and the average was 1.2 .

Triakis megalopterus (spotted gulley shark)

( fig. 24)

Two specimens of this species were included in the analysis; both from South Africa and are representative of the distribution of this species, which occurs only along the coast of South Africa. These specimens clustered together, well away from the other two species of Triakis included in the analysis. The two specimens of T. megalopterus differed by two bases. The average of the pairwise differences between T. megalopterus and T. semifasciata ( fig. 27) was 129, and the average of the pairwise differences between T. megalopterus and T. scyllium ( fig. 27) was 139.5. Differences between T. megalopterus and T. semifasciata relative to T. scyllium were noted earlier by Compagno (1988) who suggested that Triakis may constitute two subgenera: T. ( Triakis ) containing T. semifasciata and T. scyllium , and T. ( Cazon ) containing not only T. megalopterus but also T. maculata and T. acutipinna , neither of which is represented here. Our results suggest that these two clades are likely not each other’s closest relatives and thus should be considered as independent genera, rather than subgenera of Triakis .

Iago omanensis (bigeye houndshark) complex

( fig. 25)

A total of 19 specimens was included in the analysis, which yielded three clusters, each of which corresponds to the geographic origins of the included specimens: Gulf of Oman (11 specimens), the Red Sea (6 specimens), and India (2 specimens). Unfortunately, the identities of these clusters are difficult to assign to nominal taxa. At present, only two species of Iago have been formally described ( Compagno, 1984a). These are Iago omanensis , occurring from the Red Sea and Gulf of Oman to Pakistan, and Iago garricki from the western central Pacific. However, additional forms, mostly from the northeastern Indian Ocean, but also one from the Philippines, have been reported in the literature (e.g., Compagno, 1988). Based on available morphological data (i.e., images), the specimens comprising the Gulf of Oman subcluster appear to be consistent with I. omanensis , so we have used this designation for the members of this cluster. Although several additional informal names have been applied to additional forms from the Gulf of Oman (e.g., Compagno, 1988: 240, referred to ‘‘ Iago sp. nov., low fins’’; Compagno et al., 2005b, included an Iago sp. A from the Gulf of Aden to India), we are uncomfortable applying any of these names to the Indian or Red Sea clusters in the absence of additional morphological data. Thus, specimens within these clusters have been referred to as Iago cf. omanensis 1 and Iago cf. omanensis 2, respectively. The range in pairwise differences among specimens in the I. omanensis cluster was 0–8 (with an average of 3.5); the two specimens of I. cf. omanensis 1 differed by seven bases, and the range in pairwise differences among the specimens of the I. cf. omanensis 2 cluster was 0– 3 (with an average of 1.5). The averages of the pairwise differences between clusters were as follows: 12.5 (between I. omanensis and I. cf. omanensis 1), 13 (between I. omanensis and I. cf. omanensis 2), and 17.2 (between I. cf. omanensis 1 and I. cf. omanensis 2).

The haplotype map for Iago species colored by phenotype ( fig. 89A) shows no overlapping haplotypes among specimens of the three forms of Iago omanensis . The haplotype map colored by geography ( fig. 89B) confirms that haplotypes are restricted to the Gulf of Oman, the Red Sea, or India. Also shown on these maps is the clearly distinctive I. garricki from the Sulu and China Seas.

Iago garricki (longnose houndshark) ( fig. 25)

Samples from three specimens collected from the Philippines and treated by Compagno et al. (2005b), were included in the analysis (i.e., GN2224 5 JPAG 083, GN2228 5 JPAG 152, and GN4330 5 JPAG 346). These specimens clustered together, grouping most closely with the other specimens of Iago species. We have followed the identification indicated by Compagno et al. (2005b) for these specimens, who considered the Philippine specimens to be conspecific with those of I. garricki from Australia and New Caledonia. The range of pairwise differences among these specimens was 4–6 (with an average of 4.7). The average of the pairwise differences between I. garricki and those of the other Iago species are as follows: 124.5 (between I. garricki and I. omanensis ), 123.5 (between and I. garricki and I. cf. omanensis 1), and 116.7 (between I. garricki and I. cf. omanensis 2).

Galeorhinus galeus (topeshark) ( fig. 26)

Our analysis included 18 specimens of this species, which were found to comprise a single cluster. These came from South Africa (5 specimens), the Tasman Sea off Australia (1 specimen), New Zealand (7 specimens), California (3 specimens), the Azores (1 specimen), and the eastern North Atlantic (1 specimen); as such they represent much of the reported global distribution of this species, with the exception of South America and the western coast of Africa. The range in pairwise differences among all 18 specimens was 0–13, with an average of 4.9. The analysis yielded a group composed of three weak subclusters. The first subcluster consisted of the specimens from New Zealand, Australia, and South Africa with a range in pairwise differences among specimens of 0–4. The second consisted of the specimens from California with a range in pairwise differences of 0–1, and the third subcluster consisted of the specimens from Atlantic Ocean localities that were identical in sequence. At present only the single species, Galeorhinus galeus , with the type from European waters, is recognized in this genus. However, our results are consistent with those of others, for example, Compagno (1988) and Chabot and Allen (2009), who found variation in vertebral counts and genetic structure, respectively, among populations. Nonetheless, we have used the designation G. galeus for specimens from all three subclusters but note that the average of the pairwise differences between the New Zealand / Australia and California clusters is 6.7, between the New Zealand / Australia and Atlantic clusters 10.4, and between the California and Atlantic clusters 8.3. It seems likely that further investigation will result in the resurrection of synonyms of G. galeus for some of these regional subclusters.

Hypogaleus hyugaensis (blacktip topeshark)

( fig. 26)

Thirteen specimens of this species, collected and identified by Adrian Kitchingman of Western Australian Fisheries, were included in the analysis. These all came from southwestern Australia and thus represent only a small region of the distribution of this species, which also includes Japan, Taiwan, the Persian Gulf, and the east coast of Africa. The analysis yielded a single tight cluster. The range in pairwise differences within the cluster was 0–2, with an average of 0.9. These belong to what is currently considered a monotypic genus.

Hemitriakis japanica (Japanese topeshark) ( fig. 26)

In total, our analysis included nine specimens of this species collected from Japan, Taiwan, and Vietnam, thus well representing the western North Pacific distribution of this species. Many of these specimens were sent to us by Kazuhiro Nakaya from the Hokkaido University Museum (GN2597 5 HUMZ 162467, GN2598 5 HUMZ 162468, GN2599 5 HUMZ 162469, and GN2600 5 HUMZ 176993). The four specimens from Taiwan are deposited at the University of Michigan Museum of Zoology (GN1000, GN1008, GN1009, and GN1010 5 UMMZ 231964). The analysis yielded essentially a single cluster with the range in pairwise differences within the cluster being 0–7, with an average of 3.6.

Hemitriakis leucoperiptera (whitefin topeshark)

( fig. 26)

The analysis included three specimens of this species collected from the Philippines and treated by Compagno et al. (2005b). The analysis yielded a single cluster; the range in pairwise differences among these specimens was 0–1. This result may help to eliminate concerns over the identification of Philippine H. leucoperiptera raised by Compagno et al. (2005b). Their samples included a specimen (GN2232 5 JPAG 169) with a low vertebral count identified as H. leucoperiptera , a specimen (GN2225 5 JPAG 161) of unknown vertebral count and thus included by those authors among a series of specimens of uncertain identity, tentatively considered to be H. leucoperiptera , and also a specimen (GN4366 5 JPAG 139) of unknown vertebral count, tentatively identified by those authors as ‘‘ Hemitriakis cf. japanica (var PP).’’ Our results suggest that all three specimens are conspecific. These specimens grouped most closely with those of H. japanica . The average of the pairwise differences between specimens of H. leucoperiptera and those of H. japanica was 8.3.

Hemitriakis complicofasciata (ocellate topeshark)

( fig. 26)

Our analysis was based on five samples, all taken from museum specimens (GN2592 5 HUMZ 165233, GN2593 5 HUMZ 165255, GN2594 5 HUMZ 162464, GN2595 5 HU- MZ 162465 , and GN2596 5 HUMZ162466) cited by Takahashi and Nakaya (2004) in the original description of this species from the Ryuku Islands, Japan. The analysis yielded a single cluster; the range in pairwise differences among these specimens was 0–3, with an average of 1.4.

Hemitriakis falcata (sicklefin houndshark) ( fig. 26)

The single specimen of this species that was included in our analysis was taken from a specimen (GN4894 5 ANFC H 5946-01) from Western Australia. This specimen grouped most closely with those of H. complicofasciata . The average of the pairwise differences between these two species was 10.8.

Hemitriakis sp. ( fig. 26)

Interestingly, one of the specimens originally identified as Hemitriakis complicofasciata and also referenced in the original description of the species by Takahashi and Nakaya (2004), was found to group outside the cluster consisting of the specimens of H. complicofasciata and H. falcata . The average of the pairwise differences between this sample and those from the five specimens in the H. complicofasciata cluster was 29.8. The difference between this specimen and that of H. falcata was 27. The sample was taken from a specimen deposited in the Hokkaido University Museum (GN2591 5 HUMZ 165225); this specimen needs to be more closely examined in light of these findings. It is possible this specimen represents an undescribed species.

Furgaleus macki (whiskery shark) ( fig. 26)

Two specimens, both collected from Western Australia, were included in the analysis. These are fairly representative of the distribution of this species, which is known only from the western and southern regions of Australia. These two specimens comprised a single cluster ; they were identical in sequence.

Triakis scyllium (banded houndshark) ( fig. 27)

Six specimens of this species, all collected from the Izu Peninsula in Japan, and identified by Sho Tanaka, were included. The analysis yielded a single cluster with the range in pairwise differences among specimens in this cluster being 0–4, with an average of 2.1.

Triakis semifasciata (leopard shark) ( fig. 27)

In total six specimens, one from the Gulf of California and five from coastal California, were included in the analysis. These represent much of the distribution of this species. The analysis yielded a single cluster; the range in pairwise differences within the cluster was 0–1. The average of the pairwise differences between T. scyllium and T. semifasciata was 91. Given that our specimens from western California all came from the same locality, our data did not allow us to examine the differences seen by Lewallen et al. (2007) between northern and southern specimens of this species.