296 PACIFIC SCIENCE, Vol. IX, July, 1955 may best be considered as the of the FELDMANN-MAZOYER, G. 1940. Recherches sur Ptiloteae least departing from the primitive les Geramiacees de la Mediterranee occidentale. condition of the Crouanieae. 510 pp., 191 figs., 4 pk Imprimerie Min- In ~onclusion, I wish to acknowledge the erva, Alger. invaluable assistance of L. A. Garay, who - -- 1950. Sur quelques Ceramiacees de prepared the plates. Nouvelle-Zelande (suite). Paris Mus. d'Hist. North-South Differentiation of Blenniid Fishes in the Central Pacific Nat., Bul. II, 22: 307-314. KYLIN, H. 1923. Studien .iiber die Entwick- DONALD W. STRASBURGl lungsgeschichte der Florideen. Svenska Ve­ REFERENCES tensk. Akad., Handl. 63 (11): 1-138, 82 figs. --- 1930. Ueber die Entwicklungsge- DIFFERENCES in the number of fin rays, scale gether compared to previous example, but DREW, K. M. 1939. An investigation of Plu­ schichte der Florideen. Lunds Univ: Arsskr. rows, rings of bony armor (family Syngna- considerable "island-hopping" must occur if maria elegans (Bonnem.) Schmitz with spe- II, 26 (6): 1-103. thidae), and other meristic characters have there is interbreeding. cial reference to triploid plants bearing SUNESON, S. 1938. Ueber die Entwicklungs- long been employed as criteria for separating In the course of reviewing the blennioid parasporangia. Ann. Bot. [London] n.s. 3 geschichte von Plumaria elegans. Lund, species and subspecies of fishes. With the fishes of the Hawaiian Islands the writer's (10): 347-368. Fysiograftska Sdllsk., Forhandl. 8 (9): 1-8. recent increased interest in Central Pacific attention has been drawn to a similar situa- ichthyology numerous forms have been tion in that certain Hawaiian blennies bear a shown to exhibit minor differences in meristic marked similarity to others from the Marshall characters between various geographical areas, Islands. The minimum distance between these and relatively extreme populations have been two areas is about 1,400 miles, with practically described as specifically or subspecifically no islands in between. Examination of spec- distinct, depending upon the magnitude of imens from Wake Island, one of the few the differences. The term subspecies has been intermediate geographical areas, indicated a a particularly popular one to apply to slightly rather surprising degree of intermediacy with divergent populations as it connotes both respect to several morphological characters. similarity and dissimilarity. In the Pacific, Two hypotheses were formulated as a result however, its use has been confusing from the of this discovery: 1) that there might be a standpoint of zoogeography, for often little traffic of fishes between the Marshalls and or nothing has been said about gene inter- Hawaii by way of Wake, with a resultant change, effects of environmental factors, or possibility of gene interchange; and 2) that method of distribution over vast expanses of perhaps the intermediate nature of the Wake ocean. For example, Herald (in Schultz et al., specimens was due to the intermediacy of 1953 : 267, 273-275) distinguishes two sub- some factor in the physical environment. species of the pipefish Corythoichthys ftavofas­ In considering the first hypothesis it is ciatus on the basis of minor differences in noteworthy that the prevailing ocean currents meristic characters, and then says that the two affecting the Hawaiian-Marshallese area act forms are separated by a distance of 6,000 so as to move water from the Hawaiian Islands miles, with no, other representatives of the toward the Marshalls (Sverdrup et aI., 1946:

species between them. A less extreme situa- chart 7) i Consideration of this fact makes it tion is Schultz's (in Schultz et al., 1953: 292- obvious that if there is a movement of shore 297) erection of several subspecies of Atherion fishes between the Hawaiian and Marshall elymus based on slight differences in other Islands it must normally be one-way, its meristic characters. In this case the forms point of origin being in Hawaii. If this is so occur in the Philippine, Marshall, and Ma- then the high degree of endemism present in riana Islands, which are relatively close to- the Hawaiian shore fishes (reckoned as great as 52.6 per cent by Jordan and Evermann,

1 Department of Zoology, Duke University. 1905: 32) becomes a curious anomaly, as does

297 298 PACIFIC SCIENCE, Vol. IX, July, 1955 Differentiation of Blenniids - STRASBURG 299

,- WES'rERN HAWAII Northern Marshalls (Bikini, Eniwetok, S. Nat. Mus., Proc. 25 (1293): 460, 1902), Kwajalein, Rongelap) and Istiblennius zebra (Vaillant and Sauvage, 20' N ~ .. .-- u'RN HAW'" Southern Marshalls (Arno) Rev. Mag. Zoo1., 3 (3): 281,1875), while the *Line Islands (Fanning) _ .... ,\ Marquesan form is as yet undescribed. Mar- '. '.. ··.h1ARSHALLS Gilberts (Onotoa) quesan specimens were not available for this *Solomons (Bougainville, New Georgia, study, and Japanese material was not used QE-ISANDS Nissan Group) because of the difficulty of obtaining precise *East Indies (Java, Ste. Barbel) water temperatures from along the coasts of Samoa (Apia, Tutuila) Japan. Tuamotus (Fakarava, Makatea, Makemo, Principal differences between edentulus and Rangiroa) zebra are the number of fin rays, the develop- Fiji ment of the fleshy crest on the heads of 20'S Societies (Tahiti, Hereheretue) females, the presence of a tiny cirrus on each Gambiers (Mangareva) side of the nape, and the extent to which Inasmuch as the above band of islands females are covered with small dark spots. 120' 140' 160' E 180' 160'W 140' crosses the Equator it is obvious that as a Chapman (loc. cit.) admits the inconsistency FIG. 1. Map showing relationships between various island groups of the Pacific Ocean. Circles denote areas general rule sea temperatures will be greatest of the color pattern as a means of separating from which specimens were examined. near its center and least as the northern and Indian Ocean edentulus from those of the East southern extremes are approached. Factors Indies, and in the wrlter's opinion this char- the relatively disi:ant relationship between the GEOGRAPHIC AND PHYSICAL CONSIDERATIONS acting to offset this theoretical distribution acter also is not valid in the Central Pacific. fishes of the Marshalls and Hawaii, as com- are ocean currents and the seasons, each of The presence or absence of nuchal cirri has pared to the closer one between those of the The Central Pacific Ocean, using the term which is somewhat variable in itself. These been studied, and appears to be an excellent Marshalls and the East Indies. rather loosely, is crossed by a band of islands rwo factors, together with limited data on criterion for distinguishing the Hawaiian rep- In order to investigate the second hypo- running in a general southeasterly direction hydrographical conditions, make computa- resentative from the other forms of edentulus. thesis the study was restricted to a single from the Marianas to the Gambiers (Fig. 1). tion of accurate temperature values quite dif- Cirri are never present in zebra (based on 244 blenny, . . This species is This strip is about 5,500 miles in length and ficult. In this work surface water temperatures specimens), but are always present in the true widely distributed throughout the Indo- covers approximately 50 degrees (3,000 miles) have been taken from Sverdrup, et al. (1946: edentulus except for an occasional (injured?) Pacific region, but has evolved slightly differ- of latitude. Its component islands are quite charts 2 and 3), and approximations of means specimen lacking the cirrus on one side. Such ent forms in certain isolated geographical regularly spaced across the Pacific, and thus computed by averaging the values given for a loss occurs randomly throughout the dis- areas. It, together with its close relatives, is form a convenient array for a study of. the February and August. These means are listed tribution of the species, and is not restricted a tidepool inhabitant and quite demersal in effects of latitude, and hence temperature, on in Table 1, together with other data. to specimens from areas near the Hawaiian its habits. Spawning occurs in tidepools but fishes. Because of the possibility of fish move- Islands as might be expected. the postlarvae are pelagic and are undoubtedly ment between islands in this band, with re- FAUNAL CONSIDERATIONS The relative size of the cephalic crest in the stage in which dispersal of the species has sultant interbreeding and masking of tem- Istiblennius edentulus was described by Bloch females is possibly a character worthy of fur- occurred. Specimens from numerous island perature effects, it was deemed advisable to (in Bloch and Schneider, Systema ichthyolo- ther consideration. This crest is most prom- groups have been examined, and meristic data study specimens from various remote geo- giae, p. 172, 1801) from Huaheine Island in inent in zebra, and examination of large have been compared to an environmental con- graphical areas. The following list summarizes the Societies. Chapman (in de Beaufort and numbers of this blenny revealed that crest dition, water temperature, in this work. collection locales for specimens examined; Chapman, 1951: 331) records its distribution area is related to fish size but also varies The writer wishes to thank Dr. Leonard P. those regions marked with an asterisk do not from various localities throughout the Pacific somewhat at random. Insufficient material Schultz of the U. S. National Museum, for lie on or particularly near the main band of and Indian Oceans and the Red Sea. He also was available to determine the precise rela- making available the Museum's large store of islands. mentions (loc. cit.) that the forms of the spe- tionship of crest area to water temperature. Indo-Pacific blennies. Thanks are also due to *Western Hawaii (Necker, Laysan) cies occurring in southern Japan, Hawaii, and Dr. William A. Gosline and Mr. John E. the Marquesas are probably subspecifically MERISTIC DATA *Eastern Hawaii (Oahu, Maui, Hawaii) Randall, both of the University of Hawaii, distinct from the one occupying the rest of Table 1 summarizes fin ray counts made on Wake for the loan of their collections of Wake the distribution. The Japanese and Hawaiian edentulus and its close relative, zebra, for Island and Gilbert Island blennies, re- Marianas (Guam, Saipan) forms have been described as ftill species, various portions of their ranges. It also in- *Philippines (Mindoro, Balabac) spectively. Istiblennius enosimae (Jordan and Snyder, U. cludes mean counts for soft rays and data on 300 PACIFIC SCIENCE, Vol. IX, July, 1955 STRASBURG I Differentiation of Blenniids -- 301

mean surface water temperature. These data Y = 29.523 - 0.3482 X for the dorsal fin, have been arranged by latitude so that spec- and Y = 29.447 - 0.2781 X for the anal fin. imens from the northernmost islands (West- Since these two lines are approximately paral- ern Hawaii) appear at the rop, and those from lel it may be inferred that temperature has the southernmost (Gambier Islands) at the about the same effect on each fin. o 0 bottom of the table. In counting, each fin It is noteworthy that certain of the points ~ '" N N ray with a separate and distinct base was N N (solid symbols on Fig. 2) representing data listed as a single ray, and those rays split to from areas not on the main island band occur a single base were also counted as one ray. at considerable distances from the regression Inspection of Table 1 reveals a definite, but N N N lines. Fiducial limits have been set for ex- not absolutely regular, decrease in mean soft pected ray counts at various temperatures

~ N ~ ~ N ~ 0 ~ N 00 ~ ~ ~ ~ ~ ray count as one goes from northern latitudes ~ ~ N ~ N N along the regression lines, using the formula ..-< toward the Equator. Going south from the Sy = syx'\j1/n X2/ SX2 (Snedecor, 1946: . Equator the mean counts gradually increase, + ultimately approaching or exceeding those 120) and values of t at the P .Ol level. These o limits together with the actual ray counts are ,N found in the most northerly islands consid- presented in Table 2. N N 00 M N NON ~ ~ ~ ~ ~ ~ ~ 0 ered. Analysis of mean water temperatures ~ 00 M M ~ N ~ M ~ M ..-< shows that they follow the same general From Table 2 it is apparent that fin ray '">< -< pattern. counts of populatio"iis from various remote P'i o M 11"\ ~ ("("'\ C'f"I Z ~ ~ ~ ~ ~ ~ 6 8 o 0\ N M r() t<'l areas differ significantly from the values ex- ;;:: M .--I 00\0 0 0 0 ci 0\ ci ci ci ci N N N .-j N N N N N ,...... N N N N pected on the assumption of a linear rela- if> ~ ------~--~------>- in line with the expected figures. In the for- ..-< N 00 0\ ,.... 11"\ N r() "- N N N o 21.0 mer category is Eastern Hawaii (dorsal rays), I-z with Western Hawaii (anal rays) and the East o u" Indies (dorsal and anal rays) being borderline z N ..-< N ~ 20.0 0 DORSAL cases. The Line Island data are based on only ::E '" ANAL two specimens, and while their ray counts o 24.0 25.0 26.0 270 ..-< 28.0 are significantly different from the expected MEAN WATER TEMPERATURE IN values, the small sample size precludes em- ~ .--I ~ .--I 0 0 0\ N r() .--I ~ r() DEGREES CENTIGRADE 11"\ ~ .--I ,.... r() N N .--I r() ..-< phasizing them.

N FIG. 2. Relationship between mean fin ray count and mean water temperature for Istiblennius edentulus. Hollow DISCUSSION symbols represent data from main band of islands crossing Central Pacific (see text for details), solid It has been shown in Figure 2 that a close symbols denote data from outlying islands. Regression lines calculated for main island band data only. relationship exists between water temperature and fin ray count for I. edentulus. While the specimens upon which this conclusion is Figure 2 depicts the relationship between based came from a long chain of closely mean ray counts and mean water temperature, spaced islands it was also apparent (Table 2) all figures here being rounded to one decimal that specimens from certain remote islands place. Regression equations have been cal- fitted quite well into this general picture. The culated for data from the main island band Eastern Hawaiian Islands apparently represent (excluding Western and Eastern Hawaii, the a geographical region where other factors, Philippines, the Line Islands, the Solomons, presumably mutation, have offset the mean and the East Indies), and are as follows: ray count from the expected values. Western PACIFIC SCIENCE, VoL IX, ]qly, 1955 302 Differentiation of Blenniids - STRASBURG 303 TABLE 2 ACTUAL MEAN RAY COUNTS AND EXPECTED RANGE IN MEAN RAY COUNTS FOR lstiblennius edentulus for the low number of fin rays there. If the DE BEAUFORT, L. F., and W. M. CHAPMAN. AND RELATIVES FROM OUTLYING ISLANDS mutation acted to increase fin ray number 1951. The fishes of the Indo-Australian Archi­ then it would nee,d to appear at both the pelago. IX Percomorphi (concluded), Blen­ DORSAL FIN ANAL FIN northern and southern limits of the band of ISLAND noidea. xi + 484 pp. E. J. Brill, Leiden. Actual Expected Actual Expected islands, again to explain the symmetry seen DICE, L. R. 1952. Natural communities. x + count range count range in Table 1. Because of the unlikelihood of 547 pp. University of Michigan Press, Ann these occurrences, as contrasted with the Arbor. . Western Hawaii ...... : ...... 21.7 20.3- 21.9 22.2 22.2-23.2 Eastern Hawaii .. 21.9 20.3-21.5 22.3 22.1-23.0 much more plausible temperature effects, mu- JORDAN, D. S., and B. W . EVERMANN. 1905. Philippines ...... 20.1 19.6-20.3 21.8 21.6-22.0 tation may probably be ruled out from the The aquatic resources of the Hawaiian Is- Line Islands ... 20.0 20.1- 20.7 21.5 22.0-22.3 fin ray picture seen in the islands extending 1.- Solomons ...... 19.9 19.3- 20.2 21.7 21.4-22.0 lands. Part The shore fishes. U. S. Fish East Indies...... 20.2 19.3-20.2 22 .0 21.4-22.0 from the Marianas to the Gambiers. Comn., Bul. (1903) 23 (1): 1-574. It is very likely that I. zebra, the Hawaiian SCHULTZ, L. P., E. S. HERALD, E. A. LACH- form of edentulus, was derived from a stock of NER, A. D. WELANDER, and L. P. WOODS. edentulus-like blennies which acquired a high 1953. Fishes of the Marshall and Marianas Hawaii and the East Indies are borderline essentially ecotypes (as used by Dice, 1952: number of fin rays because of the temperature Islands. U. S. Nat!. Mus., Bul. 202: 1-685. instances of the same phenomenon, with the 397), but because of our lack of knowledge effects here discussed. Because of geograph- SNEDECOR, G. W. 1946. Statistical methods (ed. Line Islands possibly being so. regarding the inheritance and adaptive signifi- ical isolation and mutation it subsequently 4). xvi + 485 pp. The Iowa State College Several questions may be asked in con- cance of fin ray number this must be left as evolved into its present form. Its lack of Press, Ames, Iowa...... nection with the above statements, probably theory at present. It seems more credible that nuchal cirri, possession of a well-developed SVERDRUP, H. u., M. W. JOHNSON, and R the most important having to do with the populations from a given locale actually rep- cephalic crest in females, and distinctive fin H. FLEMING. 1946. The oceans, their physics, validity of the supposed relationship between resent ecophenes, that is, the precise number ray counts probably warrant its consideration chemistry, and general biology. x + 1087 pp. mean annual water temperature and the short of their fin rays is affected by some environ- as a full species. Prentice-Hall, Inc., New York. period in ontogeny in which the number of mental factor such as temperature. In their TANING, A. V. 1944. Experiments on meristic fin rays is determined. Clearly this presump- discussion of the response of ecophenes to and other characters in fishes. 1. On the tion can lead to certain errors, particularly in their environment neither Dice (1952: 399) REFERENCES influence of temperature on some meristic the case of fishes living in tidepools, the nor Allee et al. (1949: 62(5) mentions the characters in sea-trout and the fixation- temperature of which is notoriously variable. heritability or adaptational v~lue of th~ change ALLEE, W. c., A. E. EMERSON, O. PARK, T. period of these characters. [Denmark] Until further information is forthcoming on wrought by the response, and in light of PARK, and K. P. SCHMIDT. 1949. Principles Komm. f Havunders¢g., Meddel., Ser. Fiskeri the embryology of Central Pacific fishes, how- Taning's experimental work (1944, 1952) it of ecology. xii + 837 pp. W. B. 11 (3): 1- 66. ever, mean temperature data must be used in may probably be assumed that neither factor Saunders Company, Philadelphia and Lon- - - - 1952. Experimental study of meristic working with museum specimens. Annual is involved in the case of I. edentulus. If popu- don. characters in fishes. Bioi. Rev. 27: 169-193. data are perhaps best since they average lations within the major distribution of eden­ ~easonal vicissitudes. tulus actually represent ecophenes therr it A second question which arises has to do would be superfluous to label them with any with the taxonomic categories to be employed ' taxonomic category below the species. when dealing with a form showing meristic A third question has to do with the possible differences between various island groups. presence of mutations affecting the number Obviously, if populations from areas repre- offin rays. It may be argued that the dinal senting extreme temperatures are considered distribution of ray counts seen in Table 1 they could be regarded as sub specifically dis- could be produced by the gradual spread of tinct. When data are obtained from a more such mutations. If this were so then a muta- or less continuous series of islands, such as tion acting to decrease the number of fin rays the band running from the Marianas to the would have to occur near the Equator in order Gambiers, however, it becomes apparent that to account for the symmetrical distribution extreme populations may be united by an of ray counts on either side of that region. evenly graded series. It is possible that spec- Furthermore, it would also have to occur in imens from an island in such a barid are the Philippines and the Solomons to account