Queensland Blenny

Dorsal fin XI-XII (typically XII), 14-16 (typically 15). Anal fin II,16-18 (typically 17). Pectoral fin 12 or 13 (rarely 12). Segmented caudal-fin rays 13. Vertebrae 10 or 11 (rarely 11) + 21-24 (rarely 21, usually 23) = 31-34. Dentary incisor teeth which includes anterior canine teeth very similar in appearance from incisors, 45-53; posterior dentary canines (specimens at least 25 mm SL) 3-8 on each side (rarely 3). Lateral line lacking pairs of pores, terminating posteriorly at point between verticals from dorsal-fin spines 9 and 11 (rarely surpassing vertical from 9). With cirrus on posterior rim of anterior nostril; absent on anterior rim.

Oviparous, distinct pairing (Ref. 205).

Dorsal spines (total): 11 - 12; Dorsal soft rays (total): 14 - 16; Analspines: 2; Analsoft rays: 16 - 18

Found on coastal reefs.

Adults are found on coastal reefs (Ref. 2334). Oviparous. Eggs are demersal and adhesive (Ref. 205), and are attached to the substrate via a filamentous, adhesive pad or pedestal (Ref. 94114). Larvae are planktonic, often found in shallow, coastal waters (Ref. 94114).

Ecsenius (Ecsenius) mandibularis McCulloch

Ecsenius mandibularis McCulloch, 1923, p. 122 [Masthead Island, Australia].

DESCRIPTION.—Dorsal spines 12 (rarely 11 or 13); dorsal rays 14–16 (usually 15); segmented anal rays 16–18 (rarely 18); pectoral rays 13 (rarely 12); segmented caudal rays 13; dorsal procurrent caudal rays 6–7 (one of 53 with 8); ventral procurrent caudal rays 5–7; total caudal elements 24–27; gill-rakers 12–16 (usually 14); pseudobranchial filaments 6 (rarely 7); lower incisor teeth 45–53; lower jaw posterior canines 2–8 (usually 5–7); total lower jaw posterior canines 7–15; upper incisor teeth 120–121 (two counts); precaudal vertebrae 10 (one of 91 specimens with II); caudal vertebrae 22–23 (rarely 21 or 24); total vertebrae 31–34 (usually 32–33); epipleural ribs 12 or 13 (rarely 11 or 14). Lateral line with no paired pores, extending posteriorly to below 10th–11th dorsal spine. Dorsal fin notched seven-ninths to eight-ninths length of first dorsal ray (usually eight-ninths). Third (innermost) pelvic ray obvious to not obvious (45 of 51 specimens with ray obvious). One cirrus, rarely forked, on each anterior nostril.

Color pattern: Specimens may range from almost uniformly pale, with no distinct marks, to quite dark anteriorly, grading into moderately dusky posteriorly with two longitudinal rows of dark spots on the body. When the spots are present the upper row usually begins between the dorsal fin origin and the lateral line and terminates at some point below the dorsal fin rays. The ventral row of spots begins at about the midlateral line of the body no farther anteriorly than the level of the anus and extends to the caudal base. The upper row of spots may be obscured by the dark color of the body. I have seen collections from One Tree Island taken within a week of each other in which almost all the specimens in one collection exhibited spotting and almost none of the specimens in the other exhibited spotting. While not so restricted, the spotting seems to be found predominantly in males.

In many specimens there is a narrow, dark stripe extending from the posterior margin of the orbit across the top of the opercle. There may also be a narrow, dark stripe on each side of the underside of the head, beginning somewhat behind the tip of the lower jaw and extending to the margin of the branchiostegal membrane, thence continuous along the ventral margin of the fleshy pectoral base. The anterior margin of the lower jaw may be dark in pale as well as dark specimens. Some nonspotted specimens exhibit adumbrations of several vertical dusky bars on the upper half of the body.

The fins are not distinctively marked, usually having an overall dusky appearance. Melanophores usually concentrate along the fin elements. The caudal is darker along its midlength; the tips of the anal rays may be pale, especially in mature males; there may be a dark crescentic area at the base of the pectoral rays.

MATERIAL.—Australia, Queensland: Endeavour Reef, ANSP 109694 (2:28.1–35.2), 109692 (30.3), 109696 (14:21.4–37.2), 109698 (20.6); Little Hope Island, ANSP 109693 (4:27.3–37.1), 109695 (18.6); Big Hope Island, ANSP 109697 (29.8); Gillett Cay, Swains Reef, AMS IB–6222 (4:28.5–35.7), IB–6237 (2:15.6–17.9); Masthead Island, Capricorn Group, AMS 1–7112 (48.4, holotype of Ecsenius mandibularis, 1–7114–6 (6: 33.6–50.6); One Tree Island, Capricorn Group, USNM 201367 (2 of 56: 31.5–44.3, cleared and stained), 201821 (57:29.7–49.1); Hoskyn Island, Bunker Group, AMS IA–3585 (35.0). In addition, several series of specimens at USNM from One Tree Island were cursorily examined in preparing the color pattern description, and one specimen from USNM 201820, also One Tree Island, was used for preparation of figure 36).

ACKNOWLEDGMENTS.—For loans of specimens and other favors aiding in the completion of this study I wish to thank: F. H. Talbot and J. R. Paxton, AMS; J. E. Böhlke and J. C. Tyler, ANSP; P. H. Greenwood, A. C. Wheeler, N. B. Marshall, and G. Palmer, BMNH; J. E. Randall, BPBM; W. N. Eschmeyer and P. Sonoda, CAS; S. Rajagopalan and R. S. Lal Mohan, FMRI; H. Steinitz, HUI; K. Deckert and C. Karrer, ISZZ; E. Tortonese, MCSN; M. L. Bauchot, MNHN: W. Klausewitz, NFIS: M. Smith, RU; H. Nijssen, ZMA; I. Tomiyama, Amakusa Marine Biological Laboratory, Japan; L. Fishelson, UTAI; W. Freihofer, Stanford University.

Figures 12, 17, 20, 25, 28, and 29 were drawn by Sharon L. Chambers; Figures 13, 18, 24, 26, 27, 32, 33, 35, and 36 were drawn by Martha H. Lester; Figures 22 and 23 were drawn by Michelle R. Davis; Figures 21 was drawn by Sandra Collum; Figure 31 was drawn by Ann Schreitz; Figure 30 was drawn by Barbara Holden; Figures 14, 15, 18, 19, and 34 were drawn by Jack R. Schroeder. The graphs were prepared by Richard Goodyear.

J. A. Peters, National Museum of Natural History, was extremely helpful in advising me on computer use and programing. W. F. Smith-Vaniz, University of Miami Rosenstiel School of Marine and Atmospheric Sciences, and B. B. Collette, United States Fish and Wildlife Service, called important references to my attention. E. A. Lachner, R. H. Gibbs, Jr., S. H. Weitzman, and R. Goodyear, Division of Fishes, National Museum of Natural History, offered critical comment on portions of the manuscript and useful advice during discussions of some of the problems raised during my study.

Considerable financial and field support was provided by the staffs of the Department of Zoology and Marine Laboratory of Hebrew University, under a foreign currency grant, SFC–7–0062 (2), Drs. W. Aron and H. Steinitz, principal investigators. Much field support was provided by the staffs of the United States Navy Medical Research Unit Number 2, Captain R. H. Watten, commanding, and the Imperial Ethiopian Navy, H. I. H. Commodore Alexander Desta, commanding.

This study was aided by a grant to me from the Smithsonian Research Foundation (SG0661084).

Ecsenius mandibularis McCulloch

Ecsenius mandibularis McCulloch, 1923:122r [Masthead Island, Australia; holotype, AMS I.7112].

Ecsenius (Ecsenius) mandibularis.—Springer, 1971:39 [redescription].

DESCRIPTION.—Dorsal fin XI–XII (XII in 96% of specimens), 14–16 (modally 15). Anal fin II,16–18 (modally 17). Pectoral fin 12 or 13 (rarely 12). Segmented caudal-fin rays 13. Vertebrae 10 or 11 (11 in only 1 of 90 specimens) + 21–24 (rarely 21, modally 23) = 31–34. Dentary incisor teeth (includes anterior canine teeth, which differ little, if at all, in appearance from incisors) 45–53; posterior dentary canines (specimens at least 25 mm SL) 3–8 on each side (rarely 3). Lateral line without pairs of pores, terminating posteriorly at point between verticals from dorsal-fin spines 9 and 11 (rarely failing to reach past vertical from 9). Cirrus present on posterior rim of anterior nostril; none on anterior rim.

Preserved Color: Specimens may range from almost uniformly pale, with no distinct marks to darkly dusky anteriorly, grading into moderately dusky posteriorly (females within a collection generally less dark than males). Occasionally, there is a dark, somewhat diffuse, stripe extending posteriorly from the orbit across the head; margining the stripe ventrally, there may be a more slender, pale stripe; on each side of the ventral surface of the head, there is usually evidence of a fine, dark stripe that extends posteriorly from the lower lip to the margin of the gill membrane; the stripe is in line with a similar, fine, dark stripe that borders the ventral margin of the fleshy pectoral-fin base; this latter stripe may extend anteriorly onto the breast area covered by the gill membrane. Many specimens exhibit two longitudinal rows of widely-spaced, small, dark spots; the upper row usually begins in the area between the dorsal-fin origin and the lateral line, terminating on the upper portion of the body at some point below the segmented dorsal-fin rays; the ventral row begins just dorsal to the midline of the body and no further anteriorly than a vertical from the anus, and extends to the caudal-fin base, where the terminal spot is on the body midline. Uncommonly, nonspotted specimens may exhibit several, faint, dusky bands dorsally on the body. Dorsal fin: of males almost uniformly dusky; spinous portion of females with melanophores mostly restricted to base and along spines, segmented ray portion with melanophores in interradial membranes just suprabasally, coursing along spines, and as narrow distal edging. Anal fin dusky, paler proximally (giving appearance of pale basal stripe) and distally. Pectoral fin dusky basally and along rays. Pelvic fins dusky. Caudal fin dusky basally; more or less uniformly dusky elsewhere in males, except melanophores very sparse over rays; females similar but interradial membranes dorsally and ventrally with clear areas.

Live or Fresh Color (Plate 14: figure 5): The main difference between preserved and live coloration is that the pale postorbital stripe and spots on the body are bluish; the color is otherwise overall buff grayish. Some over-exposed color slides taken in the wild on the Great Barrier Reef show the species to be generally buff colored with two rows of white spots on the body.

SEXUAL DIMORPHISM.—Aside from color pattern described above, large, mature males tend to have longer dorsal and anal fin elements than females (Springer, 1971), and most of the caudal-fin rays of males tend to elongate.

COMPARISONS.—Within its species group, and aside from color pattern, E. mandibularis is distinct in attaining a larger size (51 mm versus 36 mm SL) and in having generally higher average numbers of segmented dorsal and anal-fin rays and caudal vertebrae than the other species. See also comparisons section under E. aequalis.

DISTRIBUTION.—E. mandibularis is restricted to the reefs on the coast of Queensland, Australia, from Haggerstone Island and Cape York Peninsula in the north to the Bunker Group reefs (due east of Gladstone) in the south.

MATERIAL (* = new material).—Australia, Queensland: Haggerstone Island, Cape York Peninsula, AMS I.20937.011* (1 specimen: 39 mm SL); Lizard Island, AMS I.18739–017* (1:29), I.21943–024* (1:35); Endeavour Reef, ANSP 109692 (1:30), I09694 (2:28, 35), 109696 (14:21–37), 109698 (1:21); Egret Reef, BPBM 31714* (1:38); Little Hope Island [near Endeavour Reef], ANSP 109693 (4:27–37), 109695 (1:19); Big Hope Island [near Endeavour Reef], ANSP 109697 (1: 30); Gillett Cay, Swains Reef, AMS IB.6222 (4:28–36), IB.6237 (2:16, 18); Masthead Island, Capricorn Group, AMS I.7112 (48, holotype of Ecsenius mandibularis), I.7114–6 (6:34–51); One Tree Island, Capricorn Group, USNM 201367 (2 of 56:32, 44, cleared and stained), 201820 (40:16–51), 201821 (57:30–49); Hoskyn Island, Bunker Group, AMS IA.3585 (1:35).

Ecsenius schroederi McKinney and Springer

Ecsenius schroederi McKinney and Springer, 1976:21 [Ambon; holotype, USNM 209743].

DESCRIPTION.—Dorsal fin XI–XIII (XII in 88% of specimens), 13 or 14 (modally 14). Anal fin II,15 or 16 (modally 16). Pectoral fin 13 or 14 (rarely 14). Segmented caudal-fin rays 13. Vertebrae 10 or 11 (11 in only 1 of 34 specimens) + 21 to 23 (22 in 86% of specimens) = 31–33. Dentary incisor teeth (includes anterior canine teeth, which differ little, if at all, in appearance from incisors) 41–53 (only 1 of 23 specimens with more than 49); posterior dentary canines (specimens at least 25 mm SL) 3–6 on each side (usually more than 3; one specimen had none on one side and 6 on the other; specimens 20–24 mm SL usually have 3 or 4). Lateral line without pairs of pores, terminating posteriorly at point between verticals from dorsal-fin spines 8 and 10 (rarely failing to reach past vertical from 8). Cirrus present on posterior rim of anterior nostril; none on anterior rim.

Preserved Color: Head dusky, occasionally darker dorsally and anteriorly; fine, dusky stripe often extending posteriorly from mid-postorbital margin to upper posterior edge of opercle; indistinct, fine, dusky stripe, often originating just posterior to corner of mouth and extending across lower cheek onto opercle; fine, dusky stripe often on ventrolateral surface of head, originating just medial to anteriormost mandibular sensory pore and extending posteriorly almost to margin of gill membranes. Body pale with three, dark, wavy pinstripes; dorsalmost stripe extends from supratemporal canal posteriorly along lateral line to point below segmented-ray portion of dorsal fin; middle stripe is continuation of mid-postorbital stripe (if present), dips ventrally in region covered by appressed pectoral fin and continues to, or almost to, caudal-fin base; ventral stripe originates at dorsalmost point of pectoral-fin axil and extends posteriorly to, or almost to, caudal-fin base; stripes may be variously interrupted; a few indistinct, pale spots may be present between ventral two stripes. Dorsal fin: spinous portion with strip of melanophores basally, above which is slender immaculate area from which sparsely distributed melanophores extend dorsally along spines; segmented-ray portion without melanophores basally, but with melanophores extending up rays. Anal fin generally dusky basally up to distal margin of incised interradial membranes, at which level melanophores concentrate to form diffuse stripe; ray tips paler distal to stripe. Pectoral fin dusky basally, with melanophores coursing along rays, infrequently with diffuse stripe-like marking on fleshy portion. Pelvic fins almost immaculate. Caudal fin with dusky area basally and melanophores coursing along rays.

Live Color (from color photograph taken by G.R. Allen at Rowley Shoals): Head dark gray above mid-orbital level, abruptly paler gray below, with fine, white stripe between the two areas; upper half of the eye black, lower half white; black portion with yellow spots and spokes connected to faint-yellow half ring bordering pupil dorsally. Body similar in color to dorsal portion of head, with row of five white spots along dorsal contour anteriorly and row of six, larger, white spots anteriorly between two dark, ventral-body stripes. Debelius (1986:96) reproduces a color photograph of E. schroederi in life, much the same as that described here, except that his figure shows six white spots dorsally and seven between the dark stripes.

COMPARISONS.—E. schroederi has a distinctive color pattern within its group, as well as within Ecsenius as a whole. The striped pattern on its body might be confused with that of some specimens of E. aequalis and E. kurti that have the body stripes represented as fine lines, but the latter two species are easily separable from E. schroederi. In the anterior region of the body of E. aequalis and E. kurti, the space between the lower and middle stripes and that between the middle and upper stripes are about equal in depth, or the lower space is only slightly less deep than the upper space. In E. schroederi, in the same region of the body, the depth of the lower space is about half the depth of the upper space.

DISTRIBUTION.—Known only from the Moluccas and Scott Reef and Rowley Shoals, off northwestern Australia.

MATERIAL (* = new material).—Indonesia, Moluccas: Ambon Island, USNM 209743 (holotype: 24 mm SL); Ceram, Marsegoe Bay [~3°S, 128°E], AMS I.18469–132* (1:29). Off northwestern Australia: Scott Reef, AMS I.21315–031* (1:23), I.21316–032* (1:22), I.21318–025* (3:13–14); Rowley Shoals, Clerke Reef, WAM P.27658–014* (9:23–30), P.27665–002* (7:22–34), P.27662–025* (12:22–36).

Footnote

1The position of the dorsal end of this line along the horizontal length of the opercle is variable within a species, and often difficult to assess unless a variety of species and well-preserved specimens are available for comparisons. With such material amassed for the present study, I have been able to convince skeptical colleagues of the validity of my assessments of the relative constancy of the character within species groups, particularly the Yaeyamaensis, Oculus, and Isos groups and Ecsenius fourmanoiri of the Opsifrontalis Group. In spite of this caveat, there are many specimens that are readily and unequivocally assessable for the character.

Appendix I

Speculative Discussion of the History of the Present-day Distribution of the Oculus-Yaeyamaensis Clade

I have hypothesized a sister-group relationship for the Oculus and Yaeyamaensis groups based primarily on a few shared color-pattern characters (see discussion under account of Yaeyamaensis Group). After arriving at this hypothesis, I noted not only that all of the species within each of the two component species groups are allopatric, but also that 12 of the 13 species contained in the hypothesized clade are allopatric (Figures 9 and 10). The thirteenth species, E. yaeyamaensis, is broadly sympatric with six of the eight species of the Oculus Group. Considering the overall present distributions of the two species groups, I found it instructive to examine how these distributions might have developed. Obviously, the partial sympatry of the two groups is evidence for dispersal.

I am unable to hypothesize unequivocally the interrelationships of the species within each of the two groups, which, methodologically, should precede vicariance scenarious such as I will propose. I have proposed interrelationships for some of the species of the Oculus Group, however, based on color pattern and distribution of the species (see account of Oculus Group), and I will propose others for some of the species of the Yaeyamaensis Group (see Appendix II). Based on past tectonic and sea-level changes, I will also propose possible vicariance scenarios that could have resulted in the present-day distributions of some of the species of the Oculus Group and will discuss others for some of the species of the Yaeyamaensis Group (also in Appendix II).

A summary of my scenario for arriving at present-day distributions of the Oculus and Yaeyamaensis groups is given in Figure 66. This figure should be compared with the present-day species distribution maps of the Oculus and Yaeyamaensis groups (Figures 9 and 10) when assessing the discussion that follows.

I have been unable to hypothesize the sister group of the Oculus-Yaeyamaensis clade. Lacking that and information on the distribution of the sister group, I am unable to hypothesize the vicariant event (and its timing) that preceded the evolution of the common ancestor. For purposes of discussion, I have arbitrarily set the timing of the event as 80 m.y.a. (Figure 66a; Upper Cretaceous). Although the common ancestor may have existed earlier than 80 m.y.a., I doubt that it continued to exist much more recently than 40 m.y.a., as explained below.

I propose that the overall distribution of the common ancestor of the clade included the southeastern coasts of the Asian Plate and east coast of Africa, just as the combined distribution of the clade does today. Although I show the ancestor's distribution extending northwestward into the Eurasian Tethys, such extension would not have been necessary for my scenario if ancient land masses (e.g., island chains) stretched between eastern Africa and southeast Asia. The distribution of the common ancestor probably did not include the Red Sea, which began forming between 16 and 41 m.y.a. (time varies according to different investigators; summarized in Cochran, 1983:48). A Miocene opening, about 7–25 m.y.a., for the southern Red Sea appears to be favored (Klausewitz, 1983). The distribution also probably did not extend to Australia, which was well separated from the southeast Asian and east African areas by 80 m.y.a.

India (plus Madagascar) had separated from African Gondwana long before 80 m.y.a., and slightly before 80 m.y.a. had separated from Madagascar and begun its movement north toward Asia (Norton and Sclater, 1979). It is possible, therefore, that the distribution of the common ancestor included Ceylon, which was integral with India, by the time of Figure 66a, even though I first indicate the ancestor's presence in Ceylon in Figure 66b, 60 m.y.a.

No species of the Oculus-Yaeyamaensis clade is currently known from the coast of India or Madagascar, and I have not included these two areas in the past distribution of the common ancestor of the clade. The present absence of representatives of the clade from Madagascar is possibly a collecting artifact; few reef-fish collections have been made there. Such does not appear to be the case with India, however, which has an active community of ichthyologists. A great many extant, widely distributed, Indo-Pacific species of coral-reef associated organisms appear to be absent from the peninsular coast of India (but not the coast of Madagascar). This is due to the general absence of coral reefs on all but the southernmost portion of that coast (Sewell, 1932), which is well supplied with river deltas and sandy shores. The consequences of the absence of coral reefs on the coast of India are important and will be mentioned again below.

In Figure 67a, the Indian Ocean area is depicted at about 75 m.y.a. (McKenzie and Sclater, 1973). At this time the Seychelles and Mascarene Plateau were attached to India.

In Figure 66b (~60 m.y.a.) I propose that India's continued movement north constricted the range of the common ancestor of the Oculus-Yaeyamensis clade. Current studies (Klootwijk et al., 1986) indicate that India first collided with Asia about 60 m.y.a. The initial contact was only at the northwestern corner of India. By this time the common ancestor of the clade had dispersed to Ceylon, and India-Ceylon was well separated from Madagascar and the Seychelles, a continental fragment of undecided origin. It is possible that the division of the common ancestor of the Oculus-Yaeyamaensis clade was effected by 60 m.y.a.

By 55–53 m.y.a., the suturing of India to Asia was essentially completed (Curray, et al., 1982; Klootwijk, et al., 1986) and the western Indian Ocean, if not already isolated by 60 m.y.a., was now isolated from the Indo-West Pacific. Similarly, the range of the common ancestor of the Oculus-Yaeyamaensis clade, if it was not divided by 60 m.y.a., was now divided by the Indian barrier. In Figure 66c (~40 m.y.a.) I show the hypothetical distribution of the Oculus and Yaeyamaensis groups after divergence. The open area between India and Asia represents the unknown portion of India that has subducted and is subducting Asia (Audley-Charles, Hurley, and Smith, 1981). Marine forms, such as the ancestor of the Oculus-Yaeyamaensis clade, in that area would have been annihilated (finding a fossil Ecsenius in the area would be congruent with my scenario, although it would probably be impossible to assign such a fossil to the Oculus-Yaeyamaensis clade with confidence). At 40 m.y.a., I propose that populations of the Yaeyamaensis Group existed on continental fragments or volcanic edifices, such as the Seychelles Bank, the Mascarene Plateau, and the Chagos-Maldive-Laccadive ridge. Although these fragments were not all connected, distances separating one fragment from the next adjacent fragment quite possibly were small (Figure 66b,c; between 39 and 35 m.y.a.; McKenzie and Sclater, 1973; Norton and Sclater, 1979). The Laccadive Islands at the northern end of the ridge, where the Yaeyamaensis Group appears to be absent today (Jones and Kumaran, 1981), have not been collected for fishes using modern methods.

I propose that the isolating effect of the suturing of India (with the unfavorable ecology of both its coasts) to Asia is the basal cause of endemism among many littoral forms, particularly coral-reef obligates, in the western Indian Ocean. I do not mean to imply, however, that all subsequent speciation in the western Indian Ocean was initiated by this event. I mean only to suggest that present-day western Indian Ocean-western Pacific Ocean sister groups should in the main owe the initiation of their divergence to the barrier formed by the suturing of India to Asia.

Hocutt (1987) was probably the first to propose that the movement of India and its suturing to Asia were the vicariant events leading to western Indian Ocean endemism. His scenario originates during the late Jurassic, about 145 m.y.a., when Africa, India, Antarctica, and Australia were joined. At that time, what was to become much of the southeast coast of Africa was not coastal. He alluded to a Tethyan biota that, he implied, extended from what would be today's eastern Mediterranean to the Malayan region. He then proposed, sequentially, that India-Madagascar's separation from Africa, India's separation from Madagascar, and India's drift north enabled the dispersal of the Tethyan biota to the newly created south coast of Africa. From Africa, the biota dispersed [via Antarctica] to Australia. Present-day similarities [and sister groups] shared by the biotas of southern Africa and western Australia are thus explained. The later joining of India to Asia isolated the western Indian Ocean. Based on Hocutt's scenario, it is possible that the sister group of the Oculus-Yaeyamensis clade was formed as a result of the separation of Africa-India-Madagascar from Australia-Antarctica about 100–120 m.y.a.

Hocutt gave a general discussion of some of the systematic and biogeographical literature that had bearing on his hypotheses. Restricting myself to the ichthyological literature, I expand upon that discussion.

Cohen (1973), in an admittedly limited analysis of endemism among Indo-Pacific shorefishes, noted that a relatively high percentage (22%) of the fishes in the western Indian Ocean are endemic. Klausewitz (1978), evaluating only the families Chaetodontidae and Pomacanthidae, corroborated Cohen's findings. Klausewitz concluded that the eastern Indian Ocean coast of the Malay Peninsula was a part of the Indian Ocean biogeographic realm, or Indian Province, which he divided into eastern and western sections. In the sense that the two sections are different, I agree with him, but my impression is that the biota of the west coast of the Malay Peninsula is only slightly separable from that of the east coast; there are relatively few endemics on the west coast (but Ecsenius lubbocki is one of them).

Cladistic studies of Indo-Pacific fishes, which might be used to test the biogeographical hypotheses proposed above, are uncommon. Smith-Vaniz' (1976) cladogram of the three species of the blenniid subgenus Musgravius (genus Plagiotremus) shows a branching that separates the two western Indian Ocean species from the western and central Pacific Ocean species (the other two subgenera and all other species of Plagiotremus are either limited to the Pacific or are widely distributed throughout the Indo-Pacific).

Smith-Vaniz (1976) indicates that only three species of the blenniid subgenus Meiacanthus (genus Meiacanthus) are present in the western Indian Ocean. All but one of the remaining species of the subgenus are restricted to the western and central Pacific. The exception ranges from the Maldives east to off Bali, and Smith-Vaniz (1987) considers it to be the sister group of a Pacific species complex (group). The sister group, or groups, of the Indian Ocean species of the subgenus Meiacanthus must, therefore, be in the western Pacific.

The blenniid genus Aspidontus consists of two species (Smith-Vaniz, 1976). One, A. dussumieri, is widely distributed in the Indo-Pacific; the other, A. taeniatus, consists of two subspecies. Aspidontus t. tractus ranges from the Red Sea and western Indian Ocean to the Malayan Peninsula and Pulau Seribu Islands, in the Java sea just north of Djakarta (hence, on the fringe of western Pacific); A. t. taeniatus ranges from Viet Nam to the central Pacific. Traditionally, the isolation and subsequent divergence of these two subspecies would be attributed to a Malayan Peninsula barrier. Although this barrier, which has been transgressed by A. t. tractus, may be responsible for the divergence, I think it just as likely that the Indian barrier initiated the formation of the two subspecies and that A. t. tractus dispersed eastward (from Ceylon?) into the fringes of the Pacific. (The two subspecies of A. taeniatus are almost, if not completely separable based on color pattern; some investigators might recognize both as species.)

Vari's (1978) cladistic analysis of the genera of the Teraponidae does not indicate the existence of western Indian Ocean-western Pacific sister groups. The distributions of some of the teraponid species might indicate that such sister groups exist, but Vari's cladistic analysis did not extend below the generic level.

Winterbottom's (1985a) vicariographic analysis of the pseudochromid subfamilies Anisochrominae and Congrogadinae resulted in some unresolved relationships among the Congrogadinae, but his tentative cladogram of the species shows distinct trends with branches generally separating the western Indian Ocean genera and species from the western Pacific-eastern Indian ocean genera and species. Winterbottom hypothesized the presence of the ancestors of the anisochromin and congrogadin lines in the proto-western Indian Ocean prior to the break up of Gondwana (100–120 m.y.a.). The anisochromins (two species) are still restricted to the western Indian Ocean, but the congrogadins currently occur in both the Indian and western Pacific oceans, and, in the western Pacific, on both the India-Australian and Asian plates. To explain the distribution of the congrogadins Winterbottom (partly presaging Hocutt, 1987) proposed the following scenario.

The congrogadin ancestor dispersed (or increased its range) to include the India-Australian plate, which later separated from Gondwana, thus leading to allopatry. Three or four speciation events occurred on the India-Australian plate, which then drifted north and collided with the Asian plate. Dispersal then occurred from the India-Australian plate to the Asian plate.

Considering only the period of geologic time we both cover, Winterbottom's scenario contrasts with mine in originating Asian plate endemics subsequent to invasion from the India-Australian plate and according India no importance in isolating the eastern from the western Indian Ocean.

The scenario I am proposing (continuing below and in Appendix II, particularly for the Yaeyamaensis Group in the Indian Ocean) could be invoked to explain some of the history of the distribution patterns shown by Winterbottom's taxa. For instance, although not a coral inhabitant, Halidesmus thomaseni (Nielsen) has a disjunct distribution comprising essentially the northeastern and northwestern coasts of the Indian subcontinent (Winterbottom, 1985, fig. 4). If one can reasonably postulate that the species was distributed continuously across the coast of the Asian plate about 60 m.y.a., then the collision of India with the Asian plate could explain the present-day disjunction.

Smith-Vaniz (1976), Vari (1978), and Winterbottom (1985a) are the only cladistic studies of Indo-Pacific fishes I know, but they lend support to the western Indian Ocean-western Pacific vicariographic scenario I propose. I have not surveyed the non-ichthyological literature, but there are numerous western Indian Ocean endemic molluscs, for instance, that may well duplicate the trend shown by the fishes (see also Hocutt, 1987).

By 20 m.y.a. (Figure 66d), Australia was closely approaching the southeastern coast of the Asian Plate (Audley-Charles, et al., 1981), the Red Sea had probably opened (Cochran, 1983), and a line of island arcs (Outer Melanesian Arcs), which had begun forming about 38 m.y.a., extended along the converging margins of the Indo-Australian and Pacific plates (Kroenke, 1984). By about 25 m.y.a., this line of island arcs already included the island chains now extending from New Ireland southeast to Fiji. On Figure 66d, I indicate that the Yaeyamaensis Group had expanded its range into the Red Sea and into coastal southeast Asia, where it became sympatric with the Oculus Group. The Oculus Group is shown as having expanded its range along the Outer Melanesian Arc and down the northwest coast of Australia.

This portion of the scenario may have a weakness in that it requires an extensive eastward dispersal of the Yaeyamaensis Group, at least across the Bay of Bengal, if not around the bay's northern coast (improbable if the Gangetic delta existed). The form that dispersed would have been a Ceylonese endemic that was ancestral to E. yaeyamaensis and E. stictus (see discussion in Appendix 2). One might well question why there was no westward dispersal of the Ceylonese endemic (which would have resulted in sympatry of Yaeyamaensis Group species).

I have not extended the range of the Yaeyamaensis Group in Figure 66d to include the northwest coast of Australia because I have arbitrarily equated speciation rates in the Yaeyamaensis and Oculus groups. I believe that the extant two Oculus Group species, one (E. paroculus) occupying the Malayan Peninsula, Sumatra, and Java, and the other (E. oculatus) occupying Christmas Island and the mid-west coast of Australia, are sister species, and that their divergence occurred after the expansion of the range of their common ancestor. On the other hand, there is but a single extant Yaeyamaensis Group species sympatric with the two Oculus Group species. Because the Yaeyamaensis and Oculus groups are essentially sympatric in much of the western Pacific and have closely similar, if not identical, ecological requirements, I assume that if a Yaeyamaensis Group species had been present in Western Australia before, or contemporaneous with, the common ancestor of the two Oculus Group species, the Yaeyamaensis Group species also would have diverged recognizably. Therefore, it would have dispersed to NW Australia relatively recently, after the divergence of the Oculus Group species. (One reviewer was concerned by this proposal of “equal” evolutionary rates.)

Woodland (1986) proposed a founder-principle mechanism to explain Australia-Asian sister groups, such as E. oculatus and E. paroculus, that is compatible with Figure 66d. According to him, as Australia moved from temperate into tropical latitudes (before 20 m.y.a.), its northern edge became available for colonization by tropical species from southeast Asia. At some critical distance, there was low-level colonization of Australia by a particular species, followed by rapid evolution (adaptation) of the founder colony, and, finally, exclusion by this colony of subsequent potential colonizers of the sister species. Subsequent dispersal of Australian endemics north into the Wallace's Line zone, and similar southern dispersal into northern Australia by Asian endemics accounts for sympatry of some sister groups. Woodland also indicated the possibility that changes in sea level during the recent Cenozoic could account for the distribution and evolution of some taxa in the Wallace's Line zone. It is a subjective decision as to whether one accepts a vicariance or founder principle scenario. The main difference is that founder principle must be invoked for each pair of sister-species relationships that exhibits the same distribution pattern, whereas a vicariance scenario need be invoked but once to explain a large number of such distribution patterns.

In Figure 66d I have not extended the range of the Oculus Group to include the northeast coast of Australia. I assume that the present-day absence of the Oculus Group from northeastern Australia indicates that there was a barrier to dispersal of the Oculus Group into that area. This presents a paradoxical problem, because the Yaeyamaensis Group is represented today by a species (endemic) in northeastern Australia (Figure 66e). It might be that the barrier prevented both species groups from dispersing to northeastern Australia, but when the barrier disappeared, the Yaeyamaensis Group dispersed first. Subsequent to the first dispersal, the barrier was re-established before an Oculus Group species was able to disperse. The various interglacial periods of the recent Cenozoic could have produced such a barrier, as discussed in Appendix II.

Appendix II

Interrelationships and Biogeography of the Species of the Yaeyamaensis Group

I have been unable to hypothesize unequivocally the interspecific relationships within the Yaeyamaensis Group. At most, only one of the three types of pectoral-fin base color patterns exhibited by the group can be plesiomorphic for the group. It follows, then, that the other two patterns must be apomorphic and at least one of the two must include a pattern common to two species (either E. yaeyamaensis plus E. stictus or E. dentex plus E. nalolo, of which each pair shares a common pectoral-fin base pattern). I propose below a set of scenarios that partially account for the distributions of the species of the Yaeyamaensis Group and some of the interrelationships of the species. The scenarios are based on the hypothesized monophyly of the Oculus and Yaeyamaensis groups, the hypothesized overall vicariographic history of this clade, and details of the tectonics of the Indian Ocean area and Australia-New Guinea.

The apparently recent geological appearance of the Gulf of Aqaba (post early Miocene, not more than 20–22 m.y.a. and possibly within the past 5–10 m.y.; Eyal et al. 1981) and its cul-de-sac relationship with the Red Sea, makes it probable that E. dentex and E. nalolo had a common ancestor that was distributed in both the Red Sea and the Gulf of Aqaba (as well as the Indian Ocean).

The Gulf of Aqaba is connected to the Red Sea by an extremely narrow opening of about 6–7 km. The greatest sill depth at the opening is about 300 m. During Pleistocene glacial periods, drops in the world sea level of 100–200 m have been proposed commonly. Such drops in sea level would have restricted the opening to about 2 km or less without, at the same time, much decreasing the length and width of the Gulf (estimates based on As Suways to the Brothers, United States Defense Mapping Agency Chart 62020, 5th Edition, August, 1976).

Ideally, it would be desirable to show that the opening to the Gulf completely closed during some past period, thus guaranteeing complete isolation of all marine organisms in the Gulf of Aqaba. Although this circumstance may have existed, I do not believe it was necessary. Considering the restricted opening to the Gulf of Aqaba, all that would be necessary would be for genetic divergence of Gulf forms to occur faster than could be offset by genetic interchange with Red Sea immigrants into the Gulf (conversely, the amount of genetic interchange between species in the Red Sea and their Gulf of Aqaba conspecific immigrants would have to be so low as to be swamped, a reasonable assumption considering the much larger area of coastline available to the species in the Red Sea).

In the present scenario, the occurrence of E. dentex at Ghardaqa, Egypt, slightly southwest of the southern tip of the Sinai Peninsula, would be evidence for recent dispersal of E. dentex out of the Gulf of Aqaba.

Ecsenius dextex is differentiated from E. nalolo (and all other species of its group) primarily in having considerably more mandibular teeth (Table 20). There is a relatively short expanse of coastline between Djetta, Saudi Arabia, and Ghardaqa (northernmost Red Sea record of E. nalolo), from which no specimens of E. dentex or E. nalolo are known. If there should be a gradual change in the number of dentary incisors in specimens from along this expanse, it might be desirable to synonymize E. dentex with E. nalolo.

There are other Gulf of Aqaba-Red Sea blenniid species that show differentiation that I have not recognized nomenclaturally. I have much less material available for most of these species, but even so there is evidence of a gradual change in their morphology from north to south (see species accounts of Ecsenius gravieri and, particularly, Ecsenius frontalis). Specimens of E. frontalis tend to indicate a gradual decrease in averages for meristic characters and numbers of dentary teeth from north to south (with a reversal of the trend in the Gulf of Aden; see also: Smith-Vaniz (1976) for Meiacanthus nigrovittatus Smith-Vaniz, and Smith-Vaniz and Springer, 1971, for Mimoblennius cirrosus Smith-Vaniz and Springer and Alloblennius pictus (Lotan)).

Winterbottom (1985b) reported that a step cline exists in Haliophis guttatus (Forsskal), a congrogadid that ranges from the Gulf of Aqaba south to Madagascar. The northernmost populations are most similar to the southernmost populations. On the other hand, he noted species in several families (Chaetodontidae, Gobiidae, Pomacentridae, Labridae) that were endemic to the northern Red Sea and had their sister group in the southern Red Sea and/or Indian Ocean. Winterbottom proposed that these examples were probably indicative of a generalized tract between the northern Red Sea and the southern Red Sea plus Indian Ocean. He argued that the most plausible explanation of this tract is a vicariant event, which was undecipherable based on information available to him. He did, however, propose the possibility that the southern Red Sea was isolated from the Indian Ocean and that conditions in the southern Red Sea then might have become unfavorable to life. Present-day occurence in the southern Red Sea of sister groups of northern Red Sea taxa would be, therefore, the result of recen dispersal from the Indian Ocean (or Gulf of Aden).

Winterbottom's scenario contrasts with that which I proposed above, and which I prefer because it seems unlikely to me that conditions in the southern Red Sea would become unfavorable to life while conditions in the more restricted (lesser volume) northern Red Sea remained favorable. Hot, dry winds blow predominantly from the Wadi-al-Arabah, to the north, southward over the Gulf of Aqaba and Red Sea, causing great evaporation with resulting high salinities in the Gulf of Aqaba and northern Red Sea (note: the Dead Sea, almost uninhabitable to all life because of its high salinities, is at the north end of the Wadi-al-Arabah). The northern Red Sea, therefore, is more likely to become uninhabitable (at least through increased salinity) than is the southern Red Sea.

An alternative explanation to speciation in the Gulf of Aqaba might be that the oceanographic conditions of the Gulf, which is much cooler and saltier than the Red Sea, merely have had an ecophenotypic effect. Winterbottom argued strongly against this possibility, and I essentially agree with him. A solution, if there is only one, to this problem will require much more material from the middle and northern coastlines of the Red Sea than is currently available.

The circumstances that might have resulted in the formation of a yaeyamaensis-stictus clade are as follows. The Torres Strait (between Cape York Peninsula and New Guinea) has been opened and closed several times during the Pleistocene through changes in sea-level (Doutch, 1972). One can invoke, therefore, a land barrier to segregate a population of Ecsenius on the Queensland coast from a conspecific population on more western Australian coasts. Such isolation could result in subsequent divergence of one or both populations and might explain adequately the existence of E. stictus, except that there is no obvious reason for the ancestral Queensland population to have been (and remain) isolated from a conspecific population in eastern New Guinea, where E. yaeyamaensis occurs today: the Great Barrier Reef (GBR) continues across Torres Strait almost to New Guinea. A barrier to dispersal between Queensland and New Guinea appears to exist nevertheless.

A cursory survey of the literature evidenced several fishes (and there must be many more) that have similar distributions to that of E. stictus: Ecsenius australianus Springer, E. mandibularis McCulloch, Petroscirtes fallax Smith-Vaniz, Meiacanthus lineatus (De Vis) (all Blenniidae), Rainfordia opercularis McCulloch (Serranidae), Corythoichthys paxtoni Dawson and Doryrhamphus negrosensis malus (Whitley) (Syngnathidae), Chaetodon rainfordi McCulloch and probably Chelmon muelleri Klunzinger (Chaetodontidae), and Stegastes apicalis (De Vis) (Pomacentridae).

Possibly, ecological conditions prevent genetic interchange between the northern end of the GBR and the southeastern end of New Guinea, on the east side, and the Northern Territory, on the west side. The western half of the Gulf of Papua, northeast of the terminus of the GBR, receives drainage from many rivers and lacks coral reefs. Although the distance from the GBR to the nearest reefs to the east, which could serve as stepping-stones for dispersal, appears to be as little as 100–150 km, even shorter distances are known to isolate species in other Ecsenius species groups (for example: E. fourmanoiri from E. fijiensis; E. monoculus). The eastern Arafura Sea coast of New Guinea, just west of the northern terminus of the GBR, and the northern Gulf of Carpentaria coast along the west side of the Cape York Peninsula, also receive a large amount of river drainage and lack coral reefs. The nearest reefs to the west of the GBR are much further removed than those to the east. I hypothesize that these circumstances are major contributors to the maintenance of endemism on the GBR.

There are many species of reef fishes with distributions that are apparently unaffected by the ecological conditions that I propose. Most notably is that of Ecsenius aequalis, known only from the Great Barrier Reef and the Trobriand Islands (just north of the southeastern tip of New Guinea). I do not believe such distributions contraindicate my explanation of the GBR endemics. Endemism in marine fishes rarely exceeds 25–30% except in the largest areas. The Red Sea, which is about comparable in extent and number of species to the GBR, but with a more highly restricted interface with the rest of the marine realm, is only reported to have about 10% endemism among its fishes (Randall, 1983). The level of endemism exhibited by GBR fishes has not been estimated, but is probably less than 10%

Although I am unable to hypothesize unequivocally the sister group of the fifth and final species of the Yaeyamaensis Group, the Maldive Islands endemic E. minutus, it is instructive to examine distributions of the Yaeyamaensis Group species in the Indian Ocean in the context of the tectonic evolution of that ocean.

Figure 67a illustrates the hypothetical distribution of major land and near-surface features in the Indian Ocean area about 75 m.y.a., at which time I hypothesized above that the ancestor of the Oculus-Yaeyamaensis clade had not been split (Figure 66a). As such, the ancestor would have occupied, among other areas, the Seychelles and Mascarene Plateau, which were attached to India. Accordingly, evolution of the Yaeyamaensis Group (and Oculus Group) species would have occurred more recently than 60 m.y.a. I propose that the common ancestor of the Yaeyamaensis Group evolved between about 60 and 40 m.y.a. (Figure 66b,c).

Between about 60 and about 39 m.y.a., India's movement north resulted in: the separation of the Seychelles and Mascarene Plateau from India, the formation of the Chagos-Maldive-Laccadive Ridge, which was attached initially at its southern end to the Mascarene Plateau (Figure 67b,c), and a distancing of this ridge from India (in particular, southern India-Ceylon) by seafloor spreading between the ridge and the coast of India. Schlich (1982) considers proposals that portions of this ridge represent rifted fragments of India as speculative, although he accepts the continental association of the ridge (hotspot traces, leaky transform faults, continental fragments of India, and mixed origin have all been proposed to explain the ridge; Sahni, 1982). Heezen and Tharp's World Ocean Floor chart (United States Office of Naval Research, 1977), appears to me to indicate ocean floor spreading between India and the ridge (indications may be unclear on my much reduced reproduction of the pertinent portion of that chart, Figure 68).

Whatever the source of the ridge, it appears that a consensus would accept its closer association with the west coast of India than exists presently. In view of the occurrence of E. yaeyamaensis in Ceylon, but only other species of the Yaeyamaensis Group on the ridge and in other areas of the western Indian Ocean, I propose that the common ancestor of the yaeyamaensis-stictus clade evolved as a Ceylon endemic and as a result of the separation of the Chagos-Maldive-Laccadive Ridge from the west coast of India before 39 m.y.a. and after 60 m.y.a., based on the hypothesized time of origin of the common ancestor of the Yaeyamaensis Group (the time of the separation could be restricted further if the age of the Chagos-Maldive-Laccadive Ridge could be established).

Ecsenius minutus in the Maldives is allopatric to, and separated by a distance of about 700 km from, both the Ceylon population of E. yaeyamaensis and the closest resident population of E. nalolo, which occurs in the Chagos Islands on the Chagos-Maldive-Laccadive Ridge. Populations of E. nalolo also exist on the Mascarene Plateau and other shores of the western Indian Ocean and Red Sea. The interesting point about this distribution pattern (Figure 10) is that the Chagos Islands population of E. nalolo is approximately 1,700 km from its nearest (possible) conspecific population, on the Mascarene Plateau. Inasmuch as the Chagos Islands were formerly attached to the Mascarene Plateau (Figure 67c), one could propose that the Chagos population was carried to its present location by the seafloor spreading that split the Chagos-Maldive-Laccadive Ridge from the Mascarene Plateau beginning about 39 m.y.a. Considering the relative age of the split and great distance separating the Chagos Islands from the Mascarene Plateau, one also might expect to find evidence that the Chagos population of E. nalolo has diverged from the other Indian Ocean populations of E. nalolo. Especially so, as there is evidence that differences appear to have developed relatively recently between the Red Sea and non-Red Sea populations of E. nalolo, and I have proposed equally recent speciation events to explain the occurrence of E. dentex in the Gulf of Aqaba and E. stictus on the Great Barrier Reef. I find no such divergence, however.

The puzzling occurrence of E. nalolo in the Chagos Islands has a parallel in the distribution of E. yaeyamaensis. The population of E. yaeyamaensis at Ceylon is approximately 1,200 km from the nearest possible coral-reef locality to the east, Andaman Islands, where a conspecific population might be expected to occur. The species is unknown in the Andamans, which have not been well collected, but I predict that E. yaeyamaensis exists there. The population of E. yaeyamaensis known to be nearest to the Ceylon population is in southern Malaysia (Figure 10), and the two populations appear to be undifferentiated. The major portion of the distribution of E. yaeyamaensis is east of Ceylon, in the Pacific Ocean, and, as proposed above, this species is presumed to have reached the Pacific by dispersal. One difficulty with eastward dispersal is that the ocean currents, if such were the dispersal mechanism, in the eastern Indian Ocean-Malaysian region flow predominantly westward all year long (Wyrtki, 1961), and have done so since at least the mid-Jurassic (Haq, 1984), about 180 m.y.a. Another difficulty is that Indian Ocean populations of E. yaeyamaensis would have been isolated from conspecific Pacific populations during the same glacial periods of lowered sea level that I have hypothesized initiated divergence of the common ancestor of the yaeyamaensis-stictus clade.

If the proposed nalolo-dentex and yaeyamaensis-stictus clades are real, then E. minutus must be the sister group of one or both of them. Even narrowing the possibilities to just these three is of little help in proposing a “clean” vicariance scenario that is consistent with currently accepted historical geology. Invoking dispersal provides similarly unsatisfactory scenarios.

The solution to these problems awaits new information and insights.

Ecsenius mandibularis, also known as the many-toothed blenny, Queensland combtooth blenny or Queensland blenny in Australia, is a species of combtooth blenny in the genus Ecsenius. It is found in coral reefs in the western Pacific ocean, including the southern edge of the Great Barrier Reef. It can reach a maximum length of 7.5 centimetres. Blennies in this species feed primarily off of plants, including benthic algae and weeds.