UNIVt;RSITY OF ILLINOIS LIBRARY AT URBANACHAMPAIGN BIOLOGY APR 9lpq? 1^ lELDlANA oology '^W SERIES, NO. 38 THP LfBPARY Ursa; arval Life in the Leaves: Arboreal Tadpole Types, with Special Attention to the Morphology, Ecology, and Behavior of the Oophagous Osteopilus brunneus (Hylidae) Larva JAN 2 6 1988 Michael J. Lannoo bioil; Daniel S. Townsend 101 BURRiL Richard J. Wassersug A Contribution in Celebration of the Distinguished Scholarship of Robert F. Inger on the Occasion of His Sixty-Fifth Birthday November 30, 1987 Publication 1381 PUBLISHED BY FIELD MUSEUM OF NATURAL HISTORY Information for Contributors to Fieldiana General: Fieldiana is primarily a journal for Field Museum staff members and research associates, although manuscripts from nonaftlliatcd authors may be considered as space permits. The Journal carries a page charge of S65 per printed page or fraction thereof. Contributions from staff, research assix:iates. and invited authors will be con- 'cred for publication regardless of ability to pay page charges, but the full charge is mandatory for nonaffiliiii ihors of unsolicited manascripts. Payn^'^"' ■>* " ^'^■'^' ^'^''' of page charges qualifies a paptT •"•• --"-'i;- -< — - ■ ;. which reduces the publication time Manuscripts should be submitted to Dr. Timothy Plowman, Scientific Editor, Fieldiana, Field Museum oi Natural ! I istory, Chicago, Illinois 60605-2496, USA. Three complete copies of the text (including title page and abstract) a"i ihe illustrations should be submitted (one original copy plus two review copies which may be machine copies). manu.scripts will be considered for publication or submitted to reviewers before all materials are complete and in the hands of the Scientific Editor Text: Manuscripts must be typewritten double-spaced on .standard-weight, SVi- by 1 1-inch paper with wide marg all four sides. For papers longer than 100 manuscript pages, authors are requested to submit a "Table of Contents, a St of Illustrations," and a "List of Tables." In most cases, the text should be preceded by an "Abstract" and should conclude with "Acknowledgments" (if any) and "Literature Cited." All measurements should be in the metric system. The format and style of headings should follow those of recent issues of Fieldiana. For more detailed style informa- Chicago Manual of Style (13th ed. ), published by The University of Chicago Press, and also recent issi; Oi /■ u'liuana. In "Literature Cited," authors are encouraged to give journal and book titles in full. Where abbreviaii' desirable (e.g.. in citation of synonymies), authors consistently should follow Botanico-Periodicum-Huntianum and TL-2 TiLXonomic Literature by F. A. Stafleu & R. S. Cowan (1976^/5^^.) (botanical papf'^ ■^'- *>■■■' ^ - • .. * ■•■■•he Biosis Data Base ( 1983) published by the BioSciences Information Service. References should be typed in the following form: Croat, T. B. 1978. Flora of Barro Colorado Island. Stanford University Press. Stanford. Calif., 943 pp. Grl'bb, p. J.. J. R. Lloyd. andT D. Pennington 1963. A comparison of montane and lowland rain forest in Ecuador. I. The forest structure, physiognomy, and floristics. Journal of Ecology. 51: 567-601. < Langix^n. E. J. M. 1979. Yage among the Siona: Cultural patterns in visions, pp. 63-80. lu Browman. D L . anJ R Schwarz. eds.. Spirits, Shamans, and Stars. Mouton Publi.shers. The Hague, Netherlands Ml RRA. J. 1946. The historic tribes of Ecuador, pp. 785-821. In Steward, J. H., ed., Handbov,^ v., .,....,, .^,,,^,,^.„, ...v.,. Vol. 2, The Andean Civilizations. Bulletin 143. Bureau of American Ethnology. Smithsonian Institution, Washington. D SroiZH. R. G. 1981. Ferns and fern allies of Guatemala. Part II. Polypodiaceae. Fieldiana: Botany, n.s. , 6: 1-522. Illustrations: Illustrations are referred to in the text as "figures" (not as "plates"). Figures must be acconit:)anied by riie indication of scale, normally a reference bar. Statements in figure captions alone, such as are not acceptable. Captions should be typed double-spaced and consecutively. See recent issue ' "^ ' ' " '" style. Figures as submitted should, whenever practicable, be 8'/2 by II inches (22 x 28 cm) and may not exceed 11'/; by 16'/: inches (30 x 42 cm). Illustrations should be mounted on boards in the arrangement you wish to obtain in the printed work. This original set should be suitable for transmission to the printer as follows: Pen and ink drawings may be originals (preferred) or photostats; shaded drawings should be originals, but within the size limitation: and photo- stats should be high-quality, glossy, black and white prints. All illustrations should be marked on the reverse with author's name, figure number(s). and "top." Original illustrations will be returned to the author upon publication unless otherwise specified. Authors who wish to publish figures that require costly special paper or color reproduction must make prior arrangements with the Scientific Editor. s'age Proofs: Fieldiana employs a two-step correction system. Each author will normally receive a copy of the led manuscript on which deletions, additions, and changes can be made and queries answered. Only one set of page 'ofs will be sent. All desired corrections of type must be made on the single set of page proofs. Changes in page proofs (as opposed to corrections) are very expensive. Author-generated changes in page proofs can only be made if the author agrees in advance to pay for them. FIELDIANA Zoology NEW SERIES, NO. 38 Larval Life in the Leaves: Arboreal Tadpole Types, with Special Attention to the Morphology, Ecology, and Behavior of the Oophagous Osteopilus brunneus (Hylidae) Larva Michael J. Lannoo Daniel S. Townsend Richard J. Wassersug Department of Anatomy Department of Anatomy Department of Anatomy Dalhousie University Dalhousie University Dalhousie University Halifax, Nova Scotia Halifax, Nova Scotia Halifax, Nova Scotia ( anada B3H 4H7 Canada B3H 4H7 Canada B3H 4H7 Present address: Present address: Department of Anatomy Biology Department The University of Ottawa The University ofScranton Ottawa, Ontario Scranton, Pennsylvania 18510 Canada KIH 8M5 A Contribution in Celebration of the Distinguished Scholarship of Robert F. Inger on the Occasion of His Sixty-Fifth Birthday r Accepted for publication May 19, 1986 November 30, 1987 Publication 1381 ^PUBLISHED BY FIELD MUSEUM OF NATURAL HISTORY © 1987 Field Museum of Natural History ISSN 0015-0754 PRINTED IN THE UNITED STATES OF AMERICA II Table of Contents Abstract 1 Introduction 1 Literature Review 2 Methods 2 Field Observations 2 Laboratory Observations 3 Swimming 3 Aerial Respiratory Patterns 3 Anoxic Tolerance 3 Resistance to Subaerial Conditions 4 Results and Discussion 5 Ecology and Behavior 5 Eggs and Larvae 5 Adults 5 Interactions Among Life History Stages: Tadpole Diets, Oophagy, and Parental Care 6 Larval Growth and Development 8 Larval Morphology 9 External Features 9 Chondrocranial and Buccal Pump Mor- phology 9 Buccopharyngeal Surface Features 11 Ventral Features 11 Dorsal Features 11 Myology and Abdominal Anatomy .... 11 Lateral Line Topography 12 Comparison of Osteopilus septentrionalis with O. brunneus 12 Chondrocranial and Buccal Pump Mor- phology 13 ■ Buccopharyngeal Surface Features 13 t Ventral Features 13 p Dorsal Features 13 Myology and Abdominal Anatomy .... 13 Lateral Line Topography 14 Overall Comparison and the Evolution of the Osteopilus brunneus Larva 14 Larval Behavior 16 Normal Resting Posture 16 Swimming Behavior and Kinematics ... 16 Respiration Patterns 18 Anoxic Tolerance 19 Resistance to Subaerial Exposure 20 Function of the Elongated Tail in Osteopi- lus brunneus Larvae 20 Arboreal Tadpole Types 25 Group 1 26 Group 2 26 Group 3 26 Group 4 26 Groups 5 and 6 (Others?) 26 Taxonomic Considerations 27 Summary 27 Acknowledgments 28 Literature Cited 29 List of Illustrations 1 . Photographs of Hohenbergia fawcettii clusters and a single plant 4 2. Osteopilus brunneus larval develop- mental stages 6 3. Growth data for Osteopilus brunneus larvae 7 4. Scanning electron micrograph of oral re- gion in an Osteopilus brunneus larva ... 8 5. Scanning electron micrograph compari- son of lower beak in Osteopilus brun- neus and O. septentrionalis larvae 8 6. Scanning electron micrograph compari- son of denticles in Osteopilus brunneus and O. septentrionalis larvae 9 7. Scanning electron micrograph compari- son of floor and roof of mouth of Osteo- pilus brunneus and O. septentrionalis larvae 10 8. Scanning electron micrograph of front of buccal floor in Osteopilus brunneus and O. septentrionalis larvae 11 9. Scanning electron micrograph of right branchial basket in dorsal view in Os- teopilus brunneus and O. septentrionalis larvae 12 10. Scanning electron micrograph close-up of medial half of right branchial basket in an Osteopilus brunneus larva 12 1 1. Scanning electron micrograph close-up of secretory surfaces of branchial food traps in Osteopilus brunneus and O. septentrionalis larvae 13 1 2. Scanning electron micrograph of front of buccal floor in Osteopilus brunneus and O. septentrionalis larvae 14 13. Scanning electron micrograph of buccal roof in an Osteopilus brunneus larva ... 14 14. Scanning electron micrograph of neuro- masts in Osteopilus brunneus and O. septentrionalis larvae 15 15. Artist's depiction oi Osteopilus brunneus larvae in their normal resting posture . . 15 16. Traces from a representative cine film of a swimming Osteopilus brunneus larva 16 17. Plot of maximum specific amplitude at the tail tip versus swimming speed in Osteopilus brunneus larvae 17 18. Plot of tail beat frequency versus specif- ic swimming speed in Osteopilus brun- neus larvae 17 19. Plot of Froude efficiencies (jj) against swimming speed in Osteopilus brunneus larvae 18 20. Plot of aerial respiratory frequency against dissolved oxygen in Osteopilus brunneus larvae 19 2 1 . Time to death for six tadpoles each of Osteopilus brunneus, Rana, and Xeno- pus exposed to extreme aquatic hypoxia and deprived access to the water's sur- face 19 22. Time to death for four tadpoles each of Osteopilus brunneus, Rana, and Xeno- pus placed on a wet substrate out of water 20 23. Schematic diagrams of representative tadpoles and their external mouth parts for the arboreal tadpole types recog- nized in this study 25 List of Tables 1 . Summary data for sexes, sizes, and asso- ciations of Osteopilus brunneus adults ... 5 2. Fineness ratios for hylid tadpoles 21 3. Summary of the literature on the mor- phology, ecology, and behavior of arbo- real tadpoles 22-24 IV Larval Life in the Leaves: Arboreal Tadpole Types, with Special Attention to the Morphology, Ecology, and Behavior of the Oophagous Osteopilus brunneus (Hylidae) Larva Abstract I Osteopilus brunneus tadpoles were collected from bromeliads in Jamaica. Among the unusual mor- phological features of these larvae are greatly re- duced gill filters and gill filaments, enlarged mus- cles for depressing the jaws and buccal floor, and an elongate tail with reduced tail fins. Most of the unique morphological features of O. brunneus tad- poles can be accounted for by the evolution of early metamorphic onset in an otherwise gener- alized hylid larva. The normal tadpole resting pos- ture is vertical, with the head pointing upward. Larvae are obligate air-breathers and can survive for more than a day on a wet surface, out of water. When submerged, they do not pump water through their buccopharyngeal cavity, but only open their mouths to take in air and food. We suggest that gill filament and tail fin reduction serve to reduce O2 loss to the hypoxic water of bromeliads. Based on the breeding behavior of the adults and the association of tadpoles with frog eggs, we conclude that O. brunneus larvae are obligatorily oopha- gous. on unfertilized conspecific eggs. We review what is known about other aboreal tadpoles and conclude that O. brunneus represents one of at least five distinct arboreal larval types. These types correlate with tadpole diet and microhabitat dif- ferences in the arboreal environment. Based on Inger's (1958) definition of a genus, O. brunneus and O. septentrionalis are probably not congeners. Introduction . . . one of the Brown Frogs begins his song of kuk-kuk, kuk-kuk, which another takes up, and another, and another, until his friends around him have joined in the strains . . . very soon the sound increases and advances, for the thing is catching, and thousands and thousands of little throats in near-by woods join in the hubbub, until from hill to hill it passes coming on and on but always dying away in the rear as it proceeds, then after reaching and passing up, the sound of this pulsating wave of harmony— for indeed it is harmony— becomes less and less, and fain- ter and fainter. . . . Panton, describing an Osteopilus brunneus chorus, quoted in Dunn (1926). Orton (1953), in a now-classic paper, discussed and illustrated the adaptive radiation of anuran tadpoles. In her figure seven and in the text, the arboreal tadpole type is described simply as a "thin, flattened larva." While this is a common mor- phology for arboreal tadpoles, it is now known that arboreal tadpoles assume a variety of shapes and sizes. Tadpoles of the genera Theloderma, Ano- theca, and certain Hyla live in arboreal habitats, but do not fit the Orton type (Duellman, 1970; Wassersug et al., 1981). Two questions emerge: What is the range of variation in tadpole morphology within the ar- boreal habitat, and how does this morphological variation relate to microhabitat and behavioral differences? In this paper we examine aspects of the chon- drocranial morphology, cranial myology, internal oral anatomy, lateral line system, and overall growth of the arboreal bromeliad-dwelling tadpole of the Jamaican hylid Osteopilus brunneus and compare it to a currently recognized congener, O. LANNOO ET AL.: LARVAL LIFE IN LEAVES septentrionalis (Trueb & Tyler, 1974), which has a generalized pond larva. Additionally, we ex- amine several aspects of the behavior and ecology of6>. ^rMAj«eM5 tadpoles, including swimming per- formance, aerial respiratory patterns, anoxic tol- erance, and tolerance to subaerial exposure. In the field we examined the natural associations of adults, tadpoles, and eggs and give evidence for the pro- duction of trophic eggs. We also survey most of the literature on the morphology and behavioral ecology of arboreal tadpoles in other families and genera and define five arboreal types. Finally, fol- lowing the ideas of Inger (1958) on the concept of the genus, based on the great difference in the mor- phology and way of life of O. brunneus and O. septentrionalis, we conclude that these two species are not valid congeners. Literature Review Trueb and Tyler ( 1 974) reviewed the taxonomic history of Osteopilus brunneus and assigned the species to its current position based on gross adult morphology. Dunn (1926) described bromeliad dwelling and oophagy in Jamaican hylids, includ- ing O. brunneus, and illustrated the elongate tad- pole morphology and external oral features, em- phasizing in simple line drawings the reduction of keratinized mouthparts in these larvae. External oral features of Jamaican hylids were also illus- trated in Noble (1931). Orton (1944), in an un- published doctoral dissertation, described some aspects of the external morphology, jaw muscu- lature, jaw cartilages, and skeleton in O. brunneus and several other arboreal hylid larvae. She con- cluded that the arboreal type of tadpole has been converged upon in several anuran families. More observations on the tadpole morphology and be- havior of Jamaican hylids, including relative gill filament size, aerial respiration, and swimming, can be found in another unpublished doctoral the- sis by Jones ( 1 967). His major focus, however, was on trying to understand why certain frogs, includ- ing O. brunneus, concentrate the green pigment bilevirdin in their bones. Laessle (1961) described and quantified several physicochemical properties of tadpole-bearing bromeliads, including their temperature and dissolved oxygen (DO) concen- trations, and speculated on the behavior of O. brunneus tadpoles (see below). Noble (1927), in his classic study of anuran life histories, mentioned O. brunneus and discussed its larval biology in the context of anuran phylogeny. Noble (1929) later examined the morphological adaptations of cer- tain African arboreal tadpoles and used O. brun- neus as an outgroup for comparison. Shreckenberg (1956) described thyroid gland development in O. brunneus and in doing so provided general devel- opmental data. Methods We studied Osteopilus brunneus both in the field in Jamaica and in our laboratory. Field Observations Fieldwork was done during September 1 984 and May 1985 on the grounds of the Hollywell Rec- reation Centre (Jamaican Forest Department) lo- cated in Hardwar Gap at the western end of the Blue Mountains of Jamaica at 1250 m elevation. The Hollywell Centre consists of 3 to 5 ha of partly cleared parkland surrounded by montane mist forest (Asprey & Robbins, 1953). Epiphytic bromeliads are common, both in the forest and on trees in the parkland, and occur in clusters of up to 10 plants. Bromeliads were identified using Adams (1972), and identifications were confirmed at the herbarium of the University of the West Indies in Mona, Jamaica. We found O. brunneus by localizing calling males to bromeliad clusters at night. On a subsequent day, all the bromeliads in a cluster were carefully taken down and dissected and all O. brunneus eggs, tadpoles, and adults collected. We also arbitrarily chose and sampled some clusters from which we had not heard males calling. Most eggs and tad- poles were preserved in 10% neutral buffered for- malin within two hours of collection. Develop- mental series of nine clutches were obtained by sampling, at regular intervals, eggs maintained in our cabin at 20 to 25°C. Adults were measured (snout-vent length, SVL) and sexed by their size (females average 1 5 to 20 mm longer than males; Schwartz & Fowler, 1973; Trueb & Tyler, 1974), association with spontaneously shed eggs, or call- ing behavior (which males continued even after capture). About 60 live tadpoles were brought back to Dalhousie University, where we studied several aspects of their physiology and behavior. FIELDIANA: ZOOLOGY Laboratory Observations '* In the laboratory we examined and staged (ac- cording to Gosner, 1 960) a sample of 75 preserved O. brunneus eggs and 1 40 tadpoles. Tadpoles were measured (SVL and total length; tail length, TL, was obtained by subtraction), and their guts ex- amined for food. Twenty larvae at different stages were dissected to examine their stomach contents. We described the external morphology and in- ternal oral and gross gastrointestinal anatomy of O. brunneus tadpoles using both light and scanning electron microscopy. Sp)ecimens were prepared for the electron microscope by the technique used by Wassersug and Rosenberg (1979) and Wassersug and Duellman (1984). Specimens of O. septen- trionalis from Florida, first described by Duellman and Schwartz (1958), were similarly examined for comparison with O. brunneus. In the laboratory we observed the normal be- havior of 60 O. brunneus tadpoles. We also ex- amined their (1) swimming kinematics, (2) aerial respiratory responses to variations in DO level, (3) aquatic anoxic tolerance, and (4) resistance to subaerial conditions. Because of the difficulty of transporting live tadpoles from Jamaica to Canada and the need to use tadpoles of the same devel- opmental stages in our tests, sample sizes for some of our behavioral studies were small. Where variance measures are given in the text, they are in the form of standard error about the mean. Swimming— We examined the swimming per- formance of 10 O. brunneus tadpoles using high- speed cinematographic techniques. Tadpoles ranged from 14 to 28 mm total length (stages 25 to 28) and swam in still water at 21 ± TC. Films were taken with a Locam (Red Lake Corp., Camp- bell, Calif) high-speed motion picture camera at 250 f/s with a 50-mm macro lens. The camera was located about 60 cm above the floor of a 50- x 50- X 20-cm aquarium. Two 650-watt incandes- cent lamps about 40 cm from the aquarium pro- vided illumination. Thirteen swimming bouts were analyzed and examined with a Hewlett Packard image analysis system consisting of an HP9834A minicomputer and an HP9872A digitizer. We recorded variations in the following: tail beat frequencies (/"); length of the propulsive wave (X); maximum amplitude at the tail tip (A); specific amplitude al varying distances along the body; and one measure of mechanical efficiency, Froude ef- ficiency (tj). All kinematic measures were made over as large a range of swimming speeds (U, in body lengths per second, Ls ') as possible accord- ing to the method of Wassersug and Hoff (1985). Only sequences of constant or nearly constant (± 0.3 U) velocity were used in these analyses. Aerial Respiratory Patterns— Six O. brun- neus tadpoles (stages 25 to 41, SVL 5 to 14 mm) were exposed to each of five levels of DO from 1.5 to 12.2 mg/liter. The DO levels were varied by bubbling N, through aged, previously aerated lap water for various lengths of time; longer N, bubbling resulted in lower DO levels. For each trial N, was bubbled into 1.5 liters of water, 200 ml of which was poured with minimum agitation into each of six 500-ml Erlenmeyer flasks, to a depth of between 35 and 40 mm; temperature was 2 1 .0°C. One tadpole was carefully added to each flask, then each flask was capped with a rubber stopper. Tadpoles were acclimated for 20 minutes, then we counted numbers of times they surfaced to breathe air during a 30-minute p>eriod. The DO levels were measured before each trial using a modified Winkler field kit (Model 8931; Ecologic Instrument Corp., Long Island, N.Y.). This meth- od is accurate to within 0.2 mg/liter. Treatments were presented in a staggered fashion; viz., in the order 3.2, 12.2, 1.5, 10.5, and 7.5 mg/liter. Tad- poles were rested and allowed to feed for at least four hours between treatments. Anoxic Tolerance — We compared 12 O. brunneus tadpoles with equal numbers ofXenopus laevis and Rana sylvatica tadpoles. We attempted to use tadpoles of as similar developmental stage, SVL, and wet weight as possible. Osteopilus were in stages 25 to 31, SVLs ranged from 5.0 to 10.8 mm, and wet weights ranged from 0.2 to 0.4 g. Xenopus and Rana were laboratory slock that had been maintained for several weeks in aerated water and were growing normally. Xenopus tadpoles were in stages 27 to 31, had SVLs from 10.0 to 14.6 mm, and weighed 1 .3 to 1 .5 g. Rana tadpoles were in stages 25 to 27, had SVLs from 8.5 to 9.3 mm, and weighed 0.5 to 0.6 g. Twelve tadpoles of each species were divided into two groups of six larvae per species and placed in separate glass staining dishes (75 x 95 x 60 mm deep) filled with aged, previously aerated lap water al 2 1 .5°C. The six dishes were then covered with nylon mesh (8 meshes/cm) secured by rubber bands and divided into two groups of three, each containing tadpoles from one of the species. Both groups were then submerged in separate larger tanks filled to a depth of 90 mm with aged tap water. This arrangement deprived the larvae access to the water's surface to gulp air. Aquatic DO con- LANNOO ET AL.: LARVAL LIFE IN LEAVES ■U»^s^i^Si;bii'^t£^fi^^3i^^ Fig. 1 . A, One large and two small clusters of the bromeliad Hohenbergia fawcettii on a tree at the Hollywell Recreation Centre study site. The large cluster is about 4 m above the ground. B, A single H. fawcettii plant in seed. Note serrations along leaf margins. centrations in both tanks at the beginning of this experiment were 1 2.2 mg/liter. Three air lines con- nected to air stones were placed in one tank, and the water was aerated to maintain a high DO level. Initially, the second tank was not aerated, and its DO level was allowed to decrease naturally as a result of tadpole respiration. After 93 hours the DO in the unaerated tank was 3.4 mg/Iiter, the DO in the aerated tank was 10.5 mg/liter, and all tadpoles were alive. At that point we became con- cerned that starvation might become a factor in our results. To reduce this chance and shorten the experiment, we induced severe hypoxic conditions by bubbling N, through air stones in the previously unaerated tank. During the next four hours of this treatment, the DO level gradually decreased to 1 .6 mg/liter. We observed and recorded tadpole ac- tivity and viability at regular intervals until all of the Rana and Xenopus larvae were dead. Resistance to Subaerial Conditions— To ex- amine the relative resistance of O. brunneus to being out of water, we compared 1 2 tadpoles (four O. brunneus, four X. laevis, and four R. sylvaticd) in subaerial conditions and observed their survival and behavior. We attempted to standardize de- velopmental stage, SVLs, and wet weights among species, which in part accounts for the small sam- ple sizes. The O. brunneus tadpoles in this exper- iment ranged from stages 37 to 40, had SVLs of 10.5 to 14.0 mm, and weighed 0.4 to 0.9 g. Lab- oratory stock Xenopus and Rana were again used for comparison. Xenopus tadpoles ranged from stage 31 to 34, had SVLs from 12.5 to 14.2 mm, and weighed 1.2 to 1.4 g. Rana tadpoles ranged from stage 27 to 29, had SVLs from 22.0 to 23.5 mm, and weighed 1.8 to 2.4 g. We placed individual tadpoles on separate pieces of filter paper cut to fit into 50-mm diameter Petri dishes. The paper was wetted with 2 ml of aged tap water and the dishes were covered. Observa- tions began at 0900 on day 1 and were terminated at 0730 of day 3 after all tadpoles died. Tadpoles were initially observed every 1 5 minutes and then every 30 minutes, but were not observed between FIELDIANA: ZOOLOGY 2400 and 0730 of days 2 and 3. If there was doubt about whether a tadpole was still alive, it was gently prodded to induce movement. Results and Discussion Ecology and Behavior ' We sampled 23 clusters of epiphytic bromeliads containing one to 10 plants each (x = 4.3 ± 0.6 plants per cluster) and located 1 to 7 m above the ground. Fourteen of these clusters were composed of Hohenbergiafawcett a (x = 5.3 ± 0.8 plants per cluster, range 1-10) a large bromeliad with thick, stiff, spiny-marginate leaves (fig. 1). Seven clusters were composed solely of Guzmanniafawcettii (x = 1.8 ± 0.5 plants, range 1-3), a large bromeliad with flexible, nonspiny leaves. The two remaining clusters contained a mixed assemblage of H. fawcettii and Vriesia sintenisii, a moderate-sized, smooth-leaved bromeliad. Osteopilus brunnens was found exclusively in H. fawcettii; we heard males call only from clusters of that species. Of the 23 clusters that we sampled, frogs, larvae, or eggs were collected from 1 1 pure H. fawcettii clusters and one mixed cluster (x = 5.8 ± 0.8 plants per cluster, n = 12, range 2-10). Osteopilus brunneus was never associated with G. fawcettii. This biased association of O. brunneus may involve some aspect of //. fawcettii that is advantageous to the frogs, such as the spiny leaf margins or more vertical leaf orientation yielding a deeper and darker central lank, or may simply reflect a preference for larger bromeliad clusters. Indeed, at Hardwar Gap, H. fawcettii and G. faw- cettii are the predominant large bromeliad species, but only the former consistently forms large ag- gregated clusters. A preference for //. fawcettii, however, cannot hold throughout Jamaica since this plant only occurs at moderately high eleva- tions in the Blue Mountains (Adams, 1 972), a small portion of the geographic range of O. brunneus (Schwartz & Fowler, 1973). We have heard O. brunneus calling at elevations up to 1 650 m on the flanks of Blue Mountain Peak, where H. fawcettii does not occur, but were unable to locale the bro- meliads which held the calling frogs. Eggs and Larvae— Twenty-six bromeliads in 1 1 clusters contained eggs and/or tadpoles of O. brunneus. including 1 3 plants with just eggs, seven plants with eggs and tadpoles, and six plants con- taining tadpoles alone. Tadpoles and eggs were Table 1 . A summary of the sexes, sizes (SVL in mm) and associations of adult Osteopilus brunneus collected at Hardwar Gap. - Associations In same Adult Sex SVL plant In same cluster A M 48.5 I B C M M 51.5 47.5 B + C, e, t D M 50.5 e. t e E F 55.5 e, t F M 44.5 G F 59.0 e H I M M 46.0 45.5 H + I G-J, e, t J F 59.5 e, t _ K L M F 49.0 ■ 61.5 . K + L K-M. e, t M Escaped Each adult is identified on the left by a capital letter and brackets to the right; e = associated eggs, t = asso- ciated tadpoles. usually found in the central tanks of bromeliads. Tadpoles, even those in late developmental stages, were always associated with the gelatinous rem- nants of egg capsules, which made the water in these tanks highly viscous. In September 1984 we found no eggs. At that time tadpoles were present in five bromeliads of the three clusters we sampled. These tadpoles ranged from stages 25 to 44 and occurred in groups of six to 25 per bromeliad. We also collected three recently metamorphosed froglets (SVL 20 to 25 mm) from one cluster. In May 1 985, in contrast, tadpoles collected from eight plants ranged from stages 21 to 3 1 and, with one exception, co-occurred with eggs. Tadpole numbers at that time ranged from three to 36 per bromeliad. We obtained accurate counts of the numbers of eggs in 18 of 20 bromeliads, which yielded an average of 276 ± 44 eggs per bromeliad (range 20-622). The majority of eggs were devel- oping normally and ranged from stages 3 to 20. Hatching in O. brunneus occurs early, at about stage 21. Tadpoles are oophagous after stage 24 (see below); however, the number of eggs in bro- meliads with tadpoles present (x = 312 ± 100, n = 6, range 20-597) was not significantly different from that with tadpoles absent (x = 258 ± 46, n = 12, range 79-662; P > 0.2), suggesting that de- veloping eggs are not being eaten by tadpoles. Adults— We captured a total of 12 adult O. brunneus from seven bromeliad clusters (x = 7.0 ± LANNOO ET AL.: LARVAL LIFE IN LEAVES B K Fig. 2. Illustration of a premetamorphic develop- mental series of Osteopilus brunneus larvae. Develop- mental stages according to Gosner (1960). A, Stage 21; B, stage 25; C, stage 30; D-E, stage 41. All scale lines = 5 mm. 2.8 plants per cluster, range 2-10). Four adults were found alone in individual clusters, the other eight occurred in aggregations of two to four frogs per cluster (table 1). The largest aggregation, four adults, included a pair of males (H, I) that occupied a large bromeliad, and two females (G, J), each of which occupied separate plants. Another cluster, containing two males (B, C) in separate plants, was the only cluster from which we ever heard more than one male call. A third cluster contained three adults, including a male-female pair (K, L; not in amplexus) in one bromeliad and a third adult (M) that escaped from another plant during collection. Associations of adults with eggs and tadpoles are given in Table 1 . Eggs and/or tadpoles occurred in six of the seven clusters that contained adults. Five clusters that contained eggs and/or tadpoles, but no adults, were significantly smaller than those with adults present (x = 4.2 ± 1.0 plants per clus- ter, range 2-8 vs. x = 7.8 ± 1.9 plants per cluster. range 5-10; t test, P < 0.05). Hence, eggs may be deposited in smaller clusters than adults normally inhabit. However, the numbers of clutches (count- ing eggs and tadpoles as separate clutches) in clus- ters that contained adults (x = 4.3 ± 2.2, n = 6, range 2-8) was significantly greater than clusters without adults (x = 1.4, n = 5, range 1-2; P < 0.05), indicating that more clutches are deposited in the larger clusters— the ones that the frogs nor- mally inhabit— than in the smaller ones. There is no clear pattern in the association of adults and offspring in individual bromeliads. Of 10 adults that we could assign to particular bro- meliads, three frogs, a male (D) and two females (G and J), occurred in plants with eggs and/or tadpoles. The male-female pair (K, L) was in a bromeliad that contained neither eggs nor tad- poles. After capture, the pair were put together in a reconstructed bromeliad and left undisturbed for 36 hours. They did not assume amplexus and no eggs were laid. Three of the four females sponta- neously shed eggs in collecting bags from one to several days after capture. Two of them did so twice in the space of two weeks following capture. We maintained the eggs, but none of them devel- oped. Calling behavior differed markedly between September 1984 and May 1985. In May calling was intensive, with nightly choruses that occurred in intermittent bouts that seemed to pulse (Dunn, 1926; see introductory quote). Bouts of chorusing lasted approximately 30 seconds to two minutes and occurred once or twice per hour. At Hardwar Gap, a single male generally called from each clus- ter, with the exception of the two males that called from one cluster (noted above). In September call- ing was sporadic, usually by single males or neigh- boring pairs; choruses never developed. Differences between May and September in call- ing behavior and the presence of developing eggs suggest a distinct seasonality to reproduction by O. brunneus at Hardwar Gap, corroborating the observations of Dunn (1926) and Jones (1967). Interactions Among Life History Stages: Tadpole Diets, Oophagy, and Parental Care— The stomachs of stage 21 to 24 tadpoles contained detritus, similar to the allochthonous material found in the bottoms of bromeliad tanks. It is unusual for tadpoles to feed before stage 25, when the opercular flap closes over the external gills. We do not know whether this detritus is actively or passively ingested. The stomachs of stage 25 to 41 tadpoles con- tained predominantly whole frog eggs, specifically FIELDIANA: ZOOLOGY E E 0) X 50 - «>,+ = SV Length 45- o , x = Total Length a a X 40- 35- a a X X " a D B 30- a a i 25- a B 20- B X 15- 10- X °xB ll**' o o o o o + + + 5- n u — 1 "I'll 1 ' 1 ' 1 ' 1 < 1 1 ■ 1 I 1 " 1 > 1 21 23 25 27 29 31 33 35 37 39 41 43 45 Developmental Stage (Gosner, 1960) Fig. 3. Growth data for Osleopilus brunneus larvae. Circles and squares indicate individual specimens, crosses and X's indicate means for samples of five or more specimens at the same developmental stage. Total length increases faster than snout-vent (SV) length between stages 25 and 30, indicating that much of the growth of these tadpoles during earlier larval development is because of a disproportionate increase in tail length. O. brunneus eggs since no other anurans with aquatic eggs occur in the Hardwar Gap area. Of 47 tadpoles that we examined, 37 had only eggs in their stomachs; the other 10 contained only detritus. The average number of eggs in 10 stom- achs was 35 ± 1 6.0 (range 5-1 84). There appeared to be a positive relationship between number of eggs and the size and stage of tadpoles. For ex- ample, a stage 27 (SVL 7 mm) tadpole had 1 2 eggs in its stomach and one intact egg in its intestine, a stage 37 (SVL 13 mm) tadpole had 25 eggs in its stomach and none in its intestine, whereas a stage 4 1 (SVL 1 5 mm) tadpole had 1 84 eggs in its stomach and two in its intestine. The larger num- ber of eggs found in these stomachs compared with the few intestinal eggs suggests that eggs pass into the intestine a few at a time. Eggs in stomachs never appeared digested, while those in intestines were usually highly digested, suggesting that the stomach is predominantly a storage site and that most, if not all, digestion takes place in the intes- tine. Ingested eggs from tadpoles collected in both September and May always appeared undevel- oped (< stage 3). They were granular, usually light colored, and closely resembled eggs that were spontaneously shed by adult females in our col- lecting bags. We never found evidence of tadpoles eating other tadpoles. Arboreal tadpoles of several tropical frogs are known to eat eggs (Dunn, 1926, 1937; Taylor, 1954; Jones, 1967; Duellman, 1970; Wassersug et al., 1981; Weygoldt, 1 987; this study). In Dendrobates pumilio (Weygoldt, 1980), D. histrionicus (Zim- LANNOO ET AL.: LARVAL LIFE IN LEAVES Fig. 4. Scanning electron micrograph of the oral re- gion in an Osteopilus brunneus larva. Note the deficiency in the oral disk dorsally. Reduced denticle rows and oral disk characterize macrophagous arboreal tadpoles. Scale line = 1.0 mm. mermann & Zimmermann, 1981), and D. specio- sus (Jungfer, 1985) females lay unfertilized eggs specifically as food for their arboreal tadpoles. Our observations of O. brunneus, along with those of Jones (1967), suggest that trophic egg production may occur in this species as well. Four lines of evidence support this view: (1) seasonality of fer- tilized egg production, (2) aseasonality of trophic egg production, (3) tadpole stomach morphology, and (4) increasing numbers of ingested eggs with tadpole size. First, as we noted above, we found developing eggs and young larvae in bromeliads during May 1985, but only larvae during September. These observations suggest that breeding does not occur in September. This conclusion is corroborated by Jones (1967) and our own observations on calling. Second, most larvae collected in both May and September had stomachs that contained undevel- oped eggs. This suggests that unfertilized eggs are deposited during the latter part of tadpole devel- opment, past the time when adults are breeding. Third, stomach morphology indicates that this or- gan acts as a storage site, suggesting a morpholog- ical response to an abundant but discontinuous food source. Fourth, because larger, older larvae had more eggs in their stomachs than younger lar- vae, these stomachs must be periodically refilled. None of this would be possible if tadpoles were only eating fertilized eggs and breeding took place during a brief period in the spring. Furthermore, we found larvae and eggs together in bromeliads during May. The majority of eggs were developing normally and ranged from stages 3 to 19, yet the eggs contained in tadpole stomachs showed no signs of development. Hence, it does not appear that tadpoles are continuously feeding on younger, prehatching tank-mates. We also collected spon- taneously shed nonfertile eggs in our collecting bags, indicating that females have the physiolog- ical ability to periodically provision tadpoles. Larval Growth and Development A developmental series for O. brunneus is il- lustrated in Figure 2. Growth data (fig. 3) indicate that total length increases much faster than SVL from stages 25 to 30. Tail elongation occurs after Fig. 5. Scanning electron micrograph comparison of the lower beak in Osteopilus brunneus (left) and O. septen- trionalis (right) larvae. Note the V-shaped lower beak in O. brunneus in anterior view. Osteopilus septentrionalis have three lower denticle rows that are absent in O. brunneus. The large, patent glottis of O. brunneus may be readily seen between the two sides of the beak. The same region in O. septentrionalis is obscured from view by infralabial papillae. Scale line = 1 .0 mm. FIELDIANA: ZOOLOGY Fig. 6. Scanning electron micrograph comparison of denticles from upper denticle rows of Osteopilus brunneus (left) and O. septentrionalis (right) larvae. Note the short blunt denticles in O. brunneus. Scale line = 0.05 mm. hatching. This change in gross body proportions seems unusual for tadpoles which usually grow isometrically; for example, the posthatching elon- gation seen in O. brunneus does not occur in pond- dwelling Bufo valliceps (Limbaugh & Volpe, 1 957) and Rana capita larvae (Volpe, 1957). We cannot convert Gosner stages to larval age for field-collected animals because we could not maintain natural food and temperature regimes for hatchlings raised in our laboratory. In our most complete laboratory series, eggs developed from stages 1 7 and 1 8 to stage 2 1 (hatching) in 72 hours, and from stage 21 to 25 in 1 70 hours. Larval Morphology The following description is based on six Os- teopilus brunneus tadpoles. One individual each was used to describe: ( 1 ) external morphology, (2) internal anatomy and myology, (3) internal oral surface features, (4) chondrocranium and buccal pump morphology, and (5) lateral line topography. All individuals were in stage 37, except the sp)ec- imen used to describe lateral line topography, which was in stage 25. External Features— An elongate, ventrally flattened tadpole, SVL 11.1 mm; total length 41.2 mm; body widest behind eyes, 6.0 mm; width at eyes 5.9 mm; maximum depth of body 4.3 mm; eyes directed more dorsally and laterally than an- teriorly; minimum distance between eyes 2.0 mm, maximum outside diameter of eyes 4.0 mm. In dorsal view there is a long extension of the snout past the nares, snout shovel-like; nostrils average size, directed anterolaterally; intemarial distance = interocular distance. Lateral and ventral skin transparent; spiracle sinistral, patent, located on edge of flattened ven- tral surface slightly more than halfway back on the body, no elevated flap present on spiracle; anus relatively long tube extending onto ventral fin, opening at or close to midline; lateral line neu- romasts not conspicuous. Tail long and thin, 38 myotomes in tail, 46 over- all, myotomes extending forward to back of eye; no tail fin on body proper; terminal portion of tail slightly rounded; myotomes near tail tip widely separated by myosepta; maximum height of tail (5.0 mm) occurs 58% caudal to tip of snout; tail at highest point = 60% muscle, 25% dorsal fin, 15% ventral fin; fin height does not decrease as rapidly posteriorly as muscle does, so that dorsal and ventral fins are subequal in size and equal in height to muscle in last 10% of tail. Mouth ventral, wide (width 2.4 mm); upper and lower beaks present (fig. 4), lower beak strongly notched with opposing sides at 90° (fig. 5); tiny serrations on both beaks; one upper denticle row with 72 very fine, blunt denticles lacking marginal cusps (fig. 6); no lower denticle rows; no distinct oral disc but blunt, marginal papillae surrounding mouth except for dorsal one-third over upper beak, papillae large, more globose laterally. Chondrocranial and Buccal Pump Mor- phology— The chondrocranium of (9. brunneus is similar to that of other macrophagous tadpoles in that the rostral region is broad and the palato- quadrate bars are robust. The chondrocranium overall is depressed; the processus muscularis is shallow, as in Anotheca spinosa (see fig. 3 in Was- sersug & Hoff", 1 982). The ceratohyals are oriented at 45° to the long axis of the tadpole rather than LANNOO ET AL.: LARVAL LIFE IN LEAVES Fig. 7. Scanning electron micrograph of floor (top) and roof (bottom) of Osteopilus brunneus (left) and O. sep- tentrionalis (right). Note the lung buds to the left and right sides of the esophagus (top SEMs; left bud partially removed) in both specimens. Scale line = 1 .0 mm. transversely. The lateral lever arm of the cerato- hyal is long; lever arm ratio (Wassersug & HofF, 1979) equals 0.37. This is high for tadpoles in general (x for 40 species = 0.32) but average to low for obligate macrophagous larvae (e.g., ratio 0.40 for /I. spinosa). The ceratohyals are short along the rostrocaudal axis, but a large buccal floor area is nevertheless achieved because of a very large 10 FIELDIANA: ZOOLOGY hypobranchial plate. The branchial baskets are small and shallow. Using equations from the work of Wassersug and Hofr(1979), we calculate a buccal volume of 5.1 fi\ for an O. brunneus tadpole (SVL 12.7 mm). This is more than twice that predicted for a typical, generalized pond larva of comparable size (table 6 and fig. 8 in Wassersug «& Hoff, 1979). Assuming 90% contraction of the orbitohyoideus muscle during buccal floor depression, the ceratohyal of O. brunneus rotates through 27° with each buccal pump stroke, which is typical for tadpoles in gen- eral (x for 40 species = 26°). BuCXrOPHARYNGEAL SURFACE FEATURES (FIGURE 7)— Ventral Features— Mouth broad anteriorly; all buccal papillae (e.g., infralabial, lingual, buccal floor arena papillae, etc.) absent (fig. 8); about 20 distinct pustules scattered about buccal floor; glot- tis large, patent, anteriorly directed (fig. 5), fully exposed between broad medium notch of ventral velum; marginal papillae on ventral velum absent (fig. 9); secretory cells on free edge of ventral velum opening individually between squamous epithelial cells rather than into large pits; branchial baskets small and shallow; gill filters with tertiary filter folds of low density (fig. 1 0); gill filter rows number 4, 5, 7, and 7 for ceratobranchials 1 through 4, respectively; filter plates extremely shallow, un- imbricated; filter rows few; gill slits visible in dor- sal view; isolated secretory cells in branchial food traps, but secretory ridges absent (fig. 1 1); gill fil- aments absent; large esophagus. Dorsal Features— Prenarial region very broad and smooth, lacking surface projections (fig. 12); nares transversely oriented; narial valves simple, without prenarial papillae or narial valve projec- tions; median ridge absent; all other papillae com- monly found on the buccal roof of tadpoles absent, but about 20 pustules scattered about roof; secre- tory zone indistinct; no enlarged or elevated se- cretory pits; dorsal velum with smooth anterior edge continuous across midline (fig. 13); a single weakly defined pressure cushion visible on each side; ciliary groove shallow but cilia present. MyoijOGy and Abdominal Anatomy— The most conspicuous muscle seen through the ventral skin is the very large orbitohyoideus; angularis com- plex and interhyoideus muscles are also large. The ratio of cross-sectional areas for the muscles that operate the ceratohyal pump, the interhyoideus and orbitohyoideus, is 0.39, which is low com- pared with that of Anotheca spinosa (0.55) but much higher than the average ratio for other mac- rophagous larvae (cf. Satel & Wassersug, 1981). Fig. 8. Scanning electron micrograph of the front of the buccal floor of an Osteopilus brunneus larva (top) and an O. septentrionalis larva (bottom). Note that the beak is wider and more anteriorly directed in O. brun- neus than in O. septentrionalis. Note also the absence of infralabial and buccal floor papillae in O. brunneus. Scale line = 1.0 mm. [Not included here are Hyla bromeliacia and H. dendroscarta, which were erroneously considered macrophagous by Satel and Wassersug (1981), who followed Salthe and Mecham (1974) in this error.] The myotomal musculature extends to back of eye. LANNOO ET AL.: LARVAL LIFE IN LEAVES 11 Fig. 9. Scanning electron micrograph of right branchial basket in dorsal view of an Osteopilus brunneus larva (left) and an O. septentrionalis larva (right). Note the thickened, papillate-free edge of the ventral velum in O. brunneus. Note also ciliary fields in the esophagus of both species. Scale line = 1 .0 mm. Liver visible through ventral skin, massive, fill- ing about one-fourth of body cavity, three-lobed; gut tube not double-coiled as in most tadpoles; stomach sacculate; maximum diameter of empty stomach 2.1 mm, diameter increases greatly when stomach contains eggs; stomach runs obliquely across whole abdominal cavity, J-shaped, with shorter ascending arm on right side; small intestine 0.4 to 0.6 mm in diameter; majority of intestinal coils lie in lower left quadrant of body cavity but additional short switch-back coils in lower right quadrant, i.e., many places where intestine folds back on itself, but tight, multiply coiled pattern Fig. 10. Scanning electron micrograph close-up of medial half of right branchial basket of an Osteopilus brunneus larva. Note the shallow nature of the filter plates and low density of the filter mesh. Compare with SEM of O. septentrionalis in Figure 9. Scale line = 0.5 mm. absent; expanded ampullae at junction of large and small bowel; large intestine oriented obliquely from left and rostral to right and caudal across lower abdomen, visible through skin in ventral view; gut length from esophagus to anus 75.5 mm; gut length to SVL ratio 6.3. Lungs small, about one-half body length. Fat bodies in abdomen large, with six, finger-like pro- jections. Lateral Line Topography— Neuromasts sin- gular, not forming stitches (fig. 14); a total of 18 supra- and infraorbital neuromasts per side. Comparison of Osteopilus septentrionalis with O. brunneus Duellman and Schwartz (1958) describe the ex- ternal morphology of the O. septentrionalis tad- pole. Here we augment their description for pur- poses of comparison with O. brunneus. We dissected two animals. The first was at stage 37, with SVL 14.5 mm and total length 31.5 mm, which was much larger than stage 37 O. brunneus tadpoles. The second animal was equal to or slight- ly smaller than our dissected O. brunneus larvae; it was stage 29, with SVL 10.3 mm and total length 27.1 mm. The O. septentrionalis tadpole is a typical pol- lywog in most characteristics including overall body shape and intestinal organization. Lateral line neu- romasts are conspicuous. With minor exceptions, the internal oral morphology of O. septentrionalis 12 FIELDIANA: ZOOLOGY ii^li^T Fig. 1 1 . Scanning electron micrograph close-up of secretory surfaces from the branchial food traps of Osteopilus brunneus (left) and O. septentrionalis (right) larvae. Note that isolated secretory pits are present in O. brunneus, but they are not organized into either rows or ridges as in O. septentrionalis. Scale line = 0.05 mm. is virtually identical to that of generalized hylid larvae described by Wassersug (1976, 1 980). Thus, only a very abbreviated description is given here. Chondrocranial and Buccal Pump Mor- phology— The chondrocranium is that of a gen- eralized hylid larva. The buccal pump morphology of this species was considered by Wassersug and Hoff (1979) in their comparative study of the tad- pole buccal pump mechanism. The processus muscularis is of average dimensions; the cerato- hyals are oriented transversely. The lever arm ra- tio for the arms of the ceratohyal is 0.29. The buccal volume for a tadpole of 1 5.5-mm SVL giv- en in Wassersug and Hoff (1979) is 7.4 ^1, which is high for a tadpole of this size, but much less than that of O. brunneus. Buccopharyngeal Surface Features (Figure 7)— Ventral Features— Two pairs of infralabial papillae, a small, more dorsal pair and a larger, more ventral pair that cross at the midline and partially obstruct the oral orifice (figs. 7-8); two small lingual papillae; buccal floor area large; buc- cal floor arena (BFA) broad, bounded by 10 to 12 BFA papillae per side; medial notch of the ventral velum very large, with the neighboring papillae of the ventral velum folded forward (fig. 9); free velar margin with large secretory pits; branchial baskets large and deep; gill filter density high; numbers of gill filter rows are 9, 12, 16, 10 for ceratobranchials 1 through 4, respectively; branchial food traps with distinct secretory ridges (fig. 11; ridges not well preserved in this specimen); glottis large, patent, dorsally directed. Dorsal Features— Anteriorly directed, archlike ridge in prenarial area; anterior edge of narial valves pustulate; narial valve with large pustulate narial valve projection; three postnarial papillae on each side running in oblique rows from anteromedial to posterolateral; a few other pustulalions scat- tered in postnarial arena; median ridge of average size with semicircular free anteroventral edge; lat- eral ridge papillae small, about half the size of the median ridge, with irregularly sculptured margins; other buccal roof papillae absent; dorsal velum with smooth free edge, broadly divided on the midline; two large, ill-defined pressure cushions per side; ciliary groove distinct. Myology and Abdominal Anatomy— Nei- ther the orbitohyoideus nor the angularis mus- culature of these tadpoles is exceptionally large. The liver is smaller than in O. brunneus. Gut length for a stage 37 animal was 128 mm; gut length to SVL ratio, 9.2. Fat bodies with fewer finger-like projections than in O. brunneus. While it is well established that herbivorous lad- poles on the average have longer alimentary canals than related carnivorous species (Noble, 1931), it is difficult to compare the relative gut lengths for tadpoles given in different studies. The gut length to SVL ratio of 6.3 that we obtained for O. brun- neus is similar to or higher than values reported for more generalized, nonobligatory macropha- gous tadpoles in previous studies (e.g., x = 5.3, range 1-1 1 in Heyer, 1973; x = 3.6, range 1.4-8.1 in Altig & Kelly, 1974; x = 4, range 3-5 in Wil- czynska, 1 98 1 ). However, such measurements are particularly prone to artifact (Heyer, 1973). Using great care we obtained a gut length to SVL ratio LANNOO ET AL.: LARVAL LIFE IN LEAVES 13 Fig. 12. Scanning electron micrograph of the front of the buccal roofs of an Osteopilus brunneus larva (top) and an O. septentrionalis larva (bottom). Note that the mouth is broader in O. brunneus and that all prenarial and postnarial surface projections seen in O. septentrion- alis are absent in O. brunneus. Scale line = 1.0 mm. of 9.2 for O. septentrionalis, but we do not believe that the gut is exceptionally long in this tadpole. Of greater significance is the relative difference between species; the gut of O. brunneus is 32% shorter than that of comparatively sized O. sep- tentrionalis. Lateral Line Topography— Neuromasts form stitches, on average two neuromasts per stitch were present (fig. 14); 69 supra- and infraorbital stitches were present, yielding a total of about 138 neu- romasts per side, which is approximately seven to eight times more neuromasts than O. brunneus has at the same developmental stage. Overall Comparison and the Evolution of the Osteopilus brunneus Larva The internal morphological features that most readily distinguish O. brunneus from O. septen- trionalis and other generalized hylid larvae are as follows: (1) a wider buccal floor and mouth, (2) reduced buccal floor and roof papillae, (3) enlarged anteriorly directed glottis, (4) smaller and shallow- er branchial baskets with half as many gill filter rows, (5) reduced gill filaments, (6) no secretory ridges in the branchial food traps, (7) undivided dorsal velum, (8) enlarged orbitohyoideus and an- gularis musculature, (9) enlarged liver, and (10) sacculate stomach. In addition the lateral line neu- romasts are greatly reduced. Certain features, surprisingly, do not differ be- tween the two forms. For example, larvae of both species have moderate, but not exceptional, lung development and a fully exposed glottis on the buccal floor. This latter feature suggests that aerial respiration is important for O. septentrionalis, as well as for O. brunneus. Both species have rela- tively large buccal pump volumes for tadpoles of their size, O. brunneus more so than O. septen- trionalis. The vast majority of features that distinguish O. brunneus from O. septentrionalis larvae are fea- tures associated with metamorphic morphology in pond larvae. At metamorphosis in typical tad- poles: denticle rows are lost; the gill filaments and gill filters are reduced; the glottis enlarges and be- comes fully exposed on the buccal floor; buccal Fig. 13. Scanning electron micrograph of the pos- terior right comer of the buccal roof of an Osteopilus brunneus larva in ventral view. Note that the dorsal velum is completely continuous across the midline. A pressure cushion on the velum is barely present; the ciliary groove, however, is distinct. Scale line = 1 .0 mm. 14 FIELDIANA: ZOOLOGY Fig. 14. Scanning electron micrograph close-up of neuromast organs in Ostcopilus brunncus (left) and O. scpicn- trionalis (right) larvae. Note that in O. brunneus neuromasts occur singly, whereas in O. septentrionalis neuromasts occur in groups of two. Scale line = 0.1 mm. floor and roof papillae regress; and secretory rows in the branchial food traps disappear. The most remarkably "metamorphic" features of the pre- metamorphic O. brunneus tadpole are in the buc- cal pump and gut. The oblique orientation of the ceratohyals, enlarged hypobranchial plate, and re- duced ceratohyals (reduced branchial baskets) col- lectively produce in O. brunneus a branchial skel- eton virtually indistinguishable from that of a Rana tadpole at stage 43, i.e., in the middle of meta- morphosis (cf. plate 22 in de Jongh, 1968). Sim- ilarly, the alimentary tract of our O. brunneus larva is like that of Rana and Bufo larvae in the middle of metamorphosis down to the level of minor folds and bends (cf. fig. 4 in Barrington, 1946). Great attention has been given lately to how minor shifts in developmental timing produce ma- jor shifts in the phenotype of animals (Bonner, 1982). We feel that evolution of the O. brunneus larva from a more generalized hylid tadpole in- volved such a heterochronic process. Specifically we propose that most of the unique features of O. brunneus tadpoles evolved by an evolutionary shift in the onset signal of metamorphosis; a in the equations of Alberch et al. (1979). We suggest that many functional systems in the O. brunneus larva are actually beginning to metamorphose be- fore they fully develop the typical larval pattern. However, the rate of metamorphosis has, if any- thing, been decreased, making the period of meta- morphosis longer. Osteopilus brunneus does not complete metamorphosis early in an absolute sense. For most of their larval life, these tadpoles appear to arrest differentiation near the middle of meta- morphosis. A similar shift in the timing of devel- opment has been proposed for the evolution of the arboreal, oophagous Anotheca spinosa larva (Wassersug, 1980). Such a shift may be the most common way in which obligatory macrophagous larvae have evolved. Evidently such shifts are ev- olutionarily simple, considering the multiple origins of arboreal macrophagy in tadpoles. Not all features, however, fit this evolutionary scenar- io, and morphological evolution in O. brunneus cannot be the result of a single shift in develop- mental timing. The most conspicuous feature of this tadpole, its elongate tail, is a specific larval specialization that is difficult to relate to prema- ture onset of metamorphosis. Fig. 15. Artist's depiction of Osteopilus brunneus larvae in their natural resting posture, based on labo- ratory observations. LANNOO ET AL.: LARVAL LIFE IN LEAVES 15 Fig. 16. Traces from a representative cine film of Osteopilus brunneus tadpole swimming at about 10 to 12 tail beats per second. The time between each frame in milliseconds is given below each figure. The distance between the crosses is 2 cm. Note the great amount of lateral movement at both the snout and tail tip. Another morphological feature that does not fit this evolutionary scenario is the neuromasts of the lateral line system. At hatching, amphibian larvae usually have only one neuromast (termed the pri- mary neuromast) present. With larval growth in many amphibians, primary neuromasts divide to form secondary neuromasts, which are oriented parallel to each other (Lannoo, 1 985). These groups of secondary neuromasts are termed stitches (Har- ris & Milne, 1966). Unlike O. septentrionalis, the primary neuromasts of O. brunneus tadpoles do not divide to form secondary neuromasts and stitches. Because neuromasts and stitches simply degenerate at metamorphosis and are absorbed by the thickening skin, neuromast morphology in O. brunneus is typical of that in younger O. septen- trionalis. Neuromast topography may, therefore, be considered relatively retarded in O. brunneus. Larval Behavior Tadpoles in rectangular and funnel-shaped con- tainers under light and dark conditions were ob- served in the laboratory. Additional behavioral observations were made during the course of our experiments. Normal Resting Posture— Singularly or in groups, Osteopilus brunneus tadpoles in our lab- oratory were usually quiescent, even when ex- posed to extremely low DO concentrations and various food types. These tadpoles are negatively buoyant. They sometimes lay on the bottom of their tank, either on their belly or side. More com- monly, they had an unusual posture in which their bodies were above the tank bottom with their snout pointed more or less upward, supported by the caudal portion of their elongate tail (fig. 1 5). The tail tip was on the substrate, bent laterally and curved either sharply or gradually into the long axis of the body as in the letter "J", depending on the angle of the body axis to the substrate (angle varied from 0° to 90°). This posture appears to require a bubble of air in either the mouth or lungs to make the rostral end of the tadpole buoyant; tadpoles denied access to air (see below) did not show this posture. Swimming Behavior and Kinematics— A va- riety of stimuli were used to elicit swimming be- havior. Compared with either Rana or Xenopus, Osteopilus larvae responded weakly to mechanical or electrical stimuli but explosively to bright il- lumination, such as when floodlights were turned on for filming. In this situation the Osteopilus swam 16 FIELDIANA: ZOOLOGY 0> T3 3 Q. E I < .30 - .26 - .22 - .18 - •5 .14 V a (0 ,10 .06 Rana & Bufo Xenopus J L J I I L 2 Specific Swimming (L/s) 6 8 Speed, U Fig. 17. Plot of maximum specific amplitude re- corded at the tail tip against swimming speed. The circles represent data for Osteopilus brunneus. The three lines are significant log curve fits for comparable data from other species previously studied (Wassersug & Hoff, 1 985; HofTA Wassersug, 1986). The top line is for tadpoles with relatively short tails; i.e., Rana catesbeiana and Bufo americanus. The middle curve is for Rana larvae with proportionately longer tails: R. septentrionalis and R. clamitans. The lowest curve is for Xenopus laevis tadpoles. Note the relatively high maximum specific am- plitude for O. brunneus. The relationship between these variables is not statistically significant for O. brunneus. with their snouts in contact with the bottom and attempted to burrow into small depressions that were present near the corners of their tank. They rarely swam for any length of time in a straight line. In a flow tank, such as that used by Feder and Wassersug (1984), they showed little stamina. Data were collected for 1 cine sequences of O. brunneus swimming at nearly constant velocity in a straight line over a range of speeds from 1 to 5 Ls '. Considering the variety of stimuli used in an eflbrt to elicit maximum velocity, it is likely that 5 Ls' is near the maximum velocity that these lai^ae can attain. Traces from alternate frames of one representative sequence are shown in Figure 16. The time between each frame in this figure is 8 msec; the figure shows about one-half of a tail beat cycle. As Figures 16 and 17 illustrate, swimming by Osteopilus larvae is characterized by a great deal of lateral movement compared to swimming in other tadpoles of comparable size (cf. figures in Wassersug & Hoff, 1985; Hoff«& Wassersug, 1986). Maximum specific amplitude (at the tail tip) ranged from 0.20 to 0.28 L and was not correlated with swimming speed (r = -0.28, P > 0.2). Specific amplitude along the body varied from 0.09 L im- mediately behind the snout to 0.28 L at the tail tip. These values for lateral movement equal or exceed the highest values recorded for Xenopus, Rana, and Bufo. The wave pattern in the tail of these tadpoles is not particularly uniform or stable. We found, for example, no correlation of wavelength with swim- ming speed (r = -0.205, P > 0.5). Tail beat frequencies ranged from 3.8 to 23.0 s ' and, as with other species, correlated positively o c o 3 o I- Speciflc Swimming Speed, U (L/s) Fig. 18. Plot of tail beat frequency against specific swimming speed. Squares represent data for Osteopilus brunneus. The stippled area encompasses comparable data from previous studies of Rana, Bufo, and Xenopus (Wassersug & Hoff. 1985; Hoff& Wassersug, 1986). The regression line is only for O. brunneus and is U = 2.5 1 + 3.36/; r = 0.67, P < 0.05). Note the steep slope for the regression; at high sp)eeds (>4 Ls ') O. brunneus larvae beat their tails approximately twice as fast as other tad- poles. LANNOO ET AL.: LARVAL LIFE IN LEAVES 17 >« u c o o £ LU .6 - .5 J I I I I I ' I Specific Swimming Speed, U (L/s) Fig. 19. Plot of the Froude efficiencies {v) of tadpoles against swimming speed, where r] is one measure of ki- nematic or propeller efficiency (see text). Squares rep- resent individual values for single swimming sequences of Osteopilus brunneus where variation in ?? was low. Vertical bars indicate ranges for efficiencies measured during single swimming bouts for O. brunneus where variation in ?? was high. The curve is a hyperbolic fit to comparable data for other species analyzed in Wassersug and Hoff(1985) and Hoffand Wassersug (1986). The r) was significantly lower for O. brunneus than for the other species swimming in the same velocity range. with swimming speed (fig. 18; U = 2.51 + 3.36/; r = 0.66, P < 0.05); but the slope of the O. brun- neus line is much higher than that of other tadpoles swimming in the same speed range. Thus, when swimming at 5 Ls" ' , an O. brunneus tadpole must beat its tail nearly 20 s', which is about twice as fast as a Rana or Xenopus larva needs to beat its tail to achieve the same velocity. This difference is also reflected in the Froude efficiency (fig. 1 9), which is a derived measure of kinematic (propeller) efficiency [77= 1 — 0.5 (1 — U/V), where V is the velocity of the propulsive wave traveling down the body]. For the swimming sequences we recorded, O. brunneus achieved the same efficiency as that reported for Rana, Bufo, or Xenopus only twice. Froude efficiency did not correlate with swimming speed (r = -0.21; P > 0.5). Over the velocity range recorded here, Froude efficiency for O. brunneus averaged 0.60 (range 0.57-0.78), which was 20% below the average ef- ficiencies for other larvae. This difference was sig- nificant (Mann-Whitney U test, P = 0.05). Overall the kinematic data suggest that the O. brunneus larva is a light-shy, poor swimmer that (1) only swims sporadically; (2) can only achieve moderate velocities, even with very high tail beat frequencies; and (3) swims with excessive lateral motion indicative of low kinematic efficiency. Respiration Patterns— Osteopilus brunneus tadpoles do not buccal pump. There is no evidence of aquatic buccopharyngeal respiration of any sort. The tadpoles only opened their mouths to take in food or air. There was an inverse correlation between DO levels and air gulping in O. brunneus in our ex- periment (fig. 20). An average rate of 1 8.4 air gulps/ hour occurred at 1 .5 mg/liter and became reduced to zero gulps in 30 minutes at 7.5 mg/liter. This relationship was not linear; a great decrease in air gulping occurred as DO concentration increased from 3.2 to 7.5 mg/liter (fig. 20). Osteopilus brunneus tadpoles in nature probably never experience DO concentrations much higher than 5 mg/liter. The highest recorded by Laessle (1961) from O. brunneus-inhahited bromeliads was 2.3 mg/liter; Jones (1967) recorded a high of 2.7 mg/liter. At normoxic and saturated conditions in the laboratory, O. brunneus larvae show little or no aerial respiration. Because they neither pump water through their pharynx nor surface for air at high DO, their respiratory needs can be met at high DO solely by cutaneous respiration. The highest aerial respiratory rates that we re- corded for O. brunneus are, somewhat surprising- ly, not atypical for tadpoles of other species at similarly low DO concentrations (Wassersug & Seibert, 1975; Feder, 1984; Feder & Wassersug, 1984; Wassersug & Feder, 1983; Marian et al., 1 980). The linear regression given in Figure 20 has a relatively low slope. For other species, there ap- pears to be a critical DO above which tadpoles avoid surfacing for air. If there is such a critical DO for O. brunneus, it occurs between 6 and 10 mg/liter. This is a high critical concentration com- pared with most tadpoles (values for other species range from 3 to 6 mg/liter in studies by Wassersug «& Seibert, 1975; Feder, 1984; Marian etal., 1980; Feder & Wassersug, 1984) and above what O. 18 FIELDIANA: ZOOLOGY Q. Q. 60 - 56 - 52 - 48 - 44 - 40 - 36 - 32 28 H 24 20 - 16 - 12 - 8 - 4 - - Y = - 2.22 X + 23.5 r = - . 57 , p = 0.001 o oooo o 4 5 6 7 8 9 Dissolved Oxygen (mg/L) T — r 10 11 12 13 Fig. 20. Plot of the aerial respiratory frequency of (9. brunneus larvae against dissolved oxygen in milligrams per liter. A linear regression was fitted to the data, with the regression equation and significance given on the figure. brunneus larvae are likely to experience in nature. Thus, O. brunneus tadpoles at natural DO con- centrations routinely breathe air, although they can survive for long periods through cutaneous respiration alone (see below). The shallow slope but high aerial respiratory frequency at moderate DO concentrations for O. brunneus characterizes tadpoles near metamorphosis in most other stud- ies (e.g., Wassersug & Seibert, 1975) that have examined the ontogeny of aerial respiration in an- uran larvae. Thus, the respiratory behavior is con- sistent with the argument given above (in the Lar- val Morphology section) that metamorphosis begins early in this species. One other genus, Hymenochirus, has a tadpole known not to irrigate buccopharyngeal surfaces. Although these tadpoles are pond-dwelling, they are similar to O. brunneus in that they oj^en their mouths only to lake in air or live prey. They are strictly macrophagous and have even more re- duction of larval branchial morphology than O. brunneus. Anoxic Tolerance— Only after artificially low- ering DO concentrations, and at 1.7 mg/liter, did tadpoles begin to succumb to anoxia (fig. 2 1 ). All Xenopus laevis died after 1 20 minutes of exposure to hypoxic water with DO concentration reduced to 1.7 mg/liter. All Rana sylvatica died after 150 minutes of hypoxia and at a DO concentration of 1.6 mg/liter. Osteopilus brunneus lasted another 90 minutes at this DO concentration, after a total of 240 minutes of extreme aquatic hypoxia. We then terminated the experiment. Four of these six O. brunneus tadpoles soon revived when placed w . _ Z3 O SI •t—> CO ^ _>. ter, enc ter, enc o o mat evid mat evid VI (A (/i Vi t/i §8 §8 eS ^ ^ Q (U (U (U < e 6 o cu . Vs o 00 in o\ u ^ o CO o 00 X Ov S 6 •^ 2 pa S o m o o PQ CQ ^ •o •o -o n ea ea -sS ^ C ^ C ea S o C o 5 o 3 O -s -. O '^ O •'^ .^ c at: «> a t: aC 2i S — o ea — . o _, o ea ti ea o, 3 ea o. ea a 3 ca 2 o c u o a o c ™ ^^ ■ S'^a-a^s^ H 3 s .2 1 nae inos 1 3 1 1 5, 1 la mariai la wilderi otheca sp 1 >5 1 -1^ II "S>"Ss ^ :& :? :^ ^^^ § g 11, 0, X t u ^ ^ 6 y « o § "^ c -^ o « P O W r-_ 00 00 <^ 5^ —J — -^ (N rn o a 13 ea "2 <= §•2 a C — o ea a 2 o H — i — -o 00 00 -•—33 52 OO S f^ o V 00 - in r" : V- c -; o t/l (A C. <^ SI « «J I/) U CQ CQ — C o o boa « J J 00 ^.M ■2 g* CQ rs >0 o — 00 C in o o o o o o j: Si x: js JS J= P. g gg g g H H HH H H 25 >. O S21 s& 21 a o c 3 TJ CO ♦^ •o aC a n .^ (N . ■^ C V "^ C O g. 2 _eo U g «J ^ := a a a •O -O T3 2 2 2 •o c "O c "2 c c o c o c o « Rc Re Re 3 "eo alo 0.15 Q. ^J ,_. c u o " o '-' o 3 O 3 O Sl:aS:^a^ C/5 OO s J3 O 2 ^ c .y o V « Eg § ^ CO 3 ■« o. § ^. c c w o «.rf e 3 O < < 5:- o (35 < o c 5 3 8 "5 a: 5 I I ^ "3 C is* 0, OS ^ 9:2 ^ :^ -2^ a: I a; C ^ s !■ il 2 1 S I X w .2 a R 2 Rc S I I:- 3" LANNOO ET AL.. LARVAL LIFE IN LEAVES 23 •- -O 'S f2i 2 E C fc V5 O as ■ V u. C U O N.-5 c "^ c °o E 00 m — "2 c^ c « « S: 6 2 e £ =« 2 03 03 03 03 o a o a o a •a ^ n St 5 o o ■- O 3 •5 "" o A- ---C X3 -^ C ca a S Jf c e^ 3 o o O o > S; f O 11) o T Xi c u tH CO B O ;i - o o T3 -o o a «2 t: c ^ c ^ c ^ c ^ c ^ c c o c o c o c o c o C o C o ^ o •- o •- o -S o -s o -5 O • -3 O ■ rr ■o Q.C aC at at a C aC at o _- o _ o „ o — o — o — o — O CQ « a w D. CO O. CO O. cQ O. CO D, CO a u o w o O o <-> O U o O O O O s-s- S^l^'i:^ ■&- ■&^ H H H H H H H — ■ — 1 m ^^ ^-, ~-^ — O E E ^ T3 O CO J3 IS 0) fc I- H o a •o CO i = §•2 a C __ o CO & ^ 5 Si t3 -O a So ^ -«i .^ -i ^ a ? ss^s §Q o S. •§ g E c E- 5 5 ^ o e -c E •2 O CO O "• . C wo 5 <-> 9 <^ ^ r ■« ooJ^Ou fe 5 op c C CO ^^ oo m :^ Sc2^. s^ b| o Ps S u I- 2 ■*• —^ •> O) CLl 2 a a „ 3 ii 5: -p. (u K =>• = 3 ^ o ^S ^ P ^ ^ ^ T3 - ^1 P 3 >" 2 ■^ « O o ;^ Sg ^ ^ w 1 1 s ^ CO S.^ -^ -^ O ^ •^ S> *; en « O. « CO S ■^ O *j 00 o 5 ^ S^:2 o O CO S 3 w •3 ^ - g CO «/^ <= « ^ " CO -o „ g J= ^ -o CO u "» i^J 3 O ^ •& CO 'o ^ « « _, -c -5 ^ b.a-g^ p « 5o -s; o a -2 P •^ E o a £ — c u 73 lU •- 00 p - - o E — 4> O 00 fe O o "^ a C^ "o (« ♦"• . „ •;- . < P ^ « JJ -'-^ = oO CO Q «5 ■£ " E ♦-> is (U "2 ^ S ° 8 rt o 5J S a s -c t„ o ^ X o ° -^ .2 ^ ♦-' C ^- r- ■S C CO O a X > a a< E o 5 C -ii O o ^ > X ^E-S -2 p "g = X CO ;S CO ^ a; E « ES: 4) P i/> -> - - § a CO ♦^ O .« 3 O. X (u O T3 CO X <« .12 en 5 X -o « D U •P X >-^ «^ c ^ "^ X 00 CO 73 < X C X o CO ■c wi T3 4) CO CO E o .S « 4i ?^ S t! X C 4) U X - 3 X X 24 HELDIANA: ZOOLOGY FREE-LIVING ARBOREAL TADPOLE TYPES Fig. 23. Schematic diagrams of represen- tative tadp>oles and their external mouth parts for the arboreal tadpole types recognized in this study. I, Elongate tadp>oles with reduced denticle rows, represented by Osteopiliis hrun- /ifusCHylidae); II, shorter-tailed tadpoles with reduced denticle rows, represented by Ano- theca spinosa (Hylidae); III, elongate tadpoles with increased denticle rows, represented by Hyla hromeliacia (Hylidae); IV, shorter-tailed tadpoles with increased denticle rows, repre- sented by Theloderma stellatum (Rhacophor- idae). V, Merlensophryne micranolis (Bufon- idac), with its odd crown, is sufficiently strange to be considered a type by itself. VI?, Other distinctive types may also exist. Not included arc the many arboreal tadpoles that do not feed. Only types I and II are either obligatorily (type I) or extensively (type II) oophagous. n cix m <:^^ Q 3Q 9 ■ others bo adapted for buirowing either into the detritus at the bottom or through the viscous fluid con- tained in the bromeliad tanks. Last, we consider Laessle's (1961) proposal that the elongate tail of O. brunneus serves to agitate and aerate the water in bromeliad tanks. Laessle (1961, p. 515) asserted that O. brunneus tadpoles I occupied the thin layer of jelly-free water at f the surface, keeping their excessively long thin tails in constant agitation while not pro- gressing appreciably. It appeared that their abdomens were stranded in the jelly and that most of the movement was imparted to the thin layer of water. In view of the low oxygen S readings it seems that the function of the unusually long tail is to insure better aera- tion in the thin layer of water above the jelly. These views, however, conflict with the behavior of undisturbed O. brunneus larvae in our labora- tory. Laessle illustrated (his fig. 9) O. brunneus ! larvae oriented horizontally below the water sur- j face in what he believed to be the natural posture ! of these tadpoles. In contrast (our fig. 1 5), we ob- I serve tadpoles most often vertical and quiescent. A similar head-up posture has been observed by Perret ( 1 962) in Acanthixalis spinosus and in Phry- \ nohyas resinifictrix by Pybum (lab. obs.; pers. i comm.). It was this motionless posture, with a resulting low-energy expense, that probably con- ! tributed to the relative success of O. brunneus in our hypoxic and subaerial exposure experiments. The only time we observed vigorous swimming in tadpoles was when we disturbed them by turning on the 650-watt lights required for filming. We conclude that, in the process of trying to observe tadpoles in bromeliads (something we could not do without bending the leaves back and shining a flashlight into the central tank), Laessle disturbed them, and it was more likely that he observed the escape behavior of these tadpoles and not their normal posture. In light of these three not completely satisfactory explanations for tail elongation in O. brunneus (respiration, aquatic locomotion, and aeration), we propose an additional view on the function of this long tail. We submit that the tail is a static postural organ, serving to position the body, and particu- larly the head of the tadpole, nearer to the water's surface. The lungs of these tadpoles serve as buoy- ancy organs and assist in this behavior. The ad- vantage of this posture may reflect the importance of pulmonary ventilation to these animals by serv- ing to reduce the energy involved in swimming to the surface (cf. Feder, 1984; Fedcr& Moran, 1985; Kramer, 1983). Or this posture may provide in- dividuals with a competitive advantage by giving them the first opportunity to ingest eggs deposited by females at the surface. Arboreal Tadpole Types As part of this study, we have attempted to com- pare and contrast the morphology of a variety of LANNOO ET AL.: LARVAL LIFE IN LEAVES 25 arboreal tadpoles, to better understand the links between the morphology and ecology of these or- ganisms. Since Orton's (1953) study, several new arboreal tadpoles have been described, and her simple generalization of these animals as thin and flattened must be expanded. Table 3 summarizes most of what is known about arboreal larvae. It appears from this table at least five, possibly more, distinct types of arboreal tadpoles exist (fig. 23). Group 1— Elongate tadpoles with attenuate bodies; tail/body ratios > 2; denticle formula less than 2/3, usually reduced to 1/1 or less; highly reduced gill filters and gill filaments; musculo- skeletal specializations for macrophagy or ooph- agy; little or no pigment. Osteopilus brunneus and the tadpoles of the other Jamaican hylids fit into this group. The arboreal Hoplophryne, Phi- lautus, and some Dendrobates larvae also belong here. These tadpoles are primarily associated with bromeliads and leaf axils rather than tree holes. They occur in relatively small volumes of water, which are likely to have little or no primary pro- ductivity as food for tadpoles. All the larvae in this group appear to be obligatorily oophagous during at least part of their lives. We suggest, as is known for several members of the Dendrobates histrionicus species group (e.g., D. pumilio, D. spe- ciosus, and D. histrionicus), that all may depend on unfertilized, trophic eggs for food. Weygoldt (1987) has recently speculated on the evolution of tadpole attendance and oophagy in Dendrobates; unfortunately, not enough is known about other genera in this group to know if his ideas apply generally. Group 2 — Shorter tadpoles with stout bodies; tail/body ratios <2; denticle formula reduced to 2/2 or less; highly reduced gill filters and gill fil- aments; little or no pigment. Among the tadpoles that fit into this group are Hyla zetecki, H. picadoi, and Anotheca spinosa. These tadpoles are carniv- orous and macrophagous. They are primarily as- sociated with bromeliads which have relatively little free water. Conspecific frog eggs may form a major part of their diet, but the evidence for oblig- atory oophagy and trophic eggs is less well docu- mented for this group than for the first group. The aggressive, omnivorous larvae of the Dendrobates quinquevittatus species group may belong here. Group 3 — Elongate tadpoles with attenuate bodies; tail/body ratio >1.7; denticle formula greater than 2/3; little or no reduction of the in- ternal oral features associated with microphagous feeding in pond larvae (e.g., well-developed gill filters); no specializations of the cranial muscu- loskeletal system for macrophagy. In this group belong Phyllodytes species, Hyla bromeliacia, H. dendroscarta, and Mantidactylus species. There may be two different types of tad- poles here. First there are those, such as the hylids, that increase their denticle formula above the 2/3 common for pond larvae by adding one or two lower rows. The other subgroup appears to be made up of Old World forms that add two to four upper denticle rows, while retaining the traditional three lower rows. Orton (1944) recognized the sharp distinction between the elongate arboreal tadpoles of group 1 and group 3, although she did not formally define the groups. Others (e.g., Salthe & Mecham, 1974, p. 445; Satel & Wassersug, 1 98 1) have lumped the two types together, which has led to an unwar- ranted presumption about the trophic ecology of the group 3 larvae. It is clear from both mor- phology and stomach content analysis that group 3 tadpoles are neither normally oophagous nor particularly specialized for macrophagy. These tadpoles all appear to eat detritus found in their arboreal tanks. The literature suggests that they occur in very small bodies of water. Their elon- gated form appears to facilitate either insinuating themselves among leaf axils or subaerial loco- motion between leaf axils, rather than swimming per se. Group 4— Shorter tadpoles with stout bodies; tail/body ratios <2; denticle formula greater than 2/3 (typically with two to four extra rows superi- orly); little or no reduction of internal oral features associated with macrophagy; gill filters and gill filaments normal to greater than normal density; darkly pigmented. Tadpoles included in this group are rhacophor- ids in the genera Rhacophorus, Theloderma, and Nyctixalus as well as the hyperoliid Acanthixalus. This is primarily an Old World group, but not strictly so, since Phrynohyas resinifictrix evidently also belongs here. These larvae retain the internal oral morphology of tadpoles with generalized diets and aquatic, buccopharyngeal respiration. As such, they appear restricted to larger aquatic bodies, which are more likely to occur in tree holes than leaf axils. Groups 5 and 6 (Others?)— The bufonid Mer- tensophryne micranotis is so different from other tadpoles that it may define its own group although it is most similar to tadpoles of group 1 (fig. 23). This tadpole lacks early lung development, and aerial respiration would seem limited for these 26 FIELDIANA: ZOOLOGY larvae. The strange crown on the head of this tadpole (and tadpoles of the related bufonid Steph- opaedes; Channing, 1978), however, appears to keep the nostrils at or near the water's surface. This would allow for surface film respiration. Whether it also allows for aerial respiration via buccopharyngeal ventilation is not known. A variety of tadpoles, which are not unusual for the families in which they occur (e.g., species in the microhylid genera Microhyla, Chaperina, Ka- loula conjuncta, Dendrobates auratus; the lepto- dactylid Crossodactylodes spp., the dendrobatids Colostethus bromelicola, Phyllobates vittatus and P. lugubris. and hytids of the Ololygon perpusilla species group) are sometimes found in tree holes or leaf axils. Tadpoles like these may represent a morphological gradation between the generalized pond tadpole and more specialized arboreal lad- poles, such as those of group 2 above. Among Dendrobates there appears to be a complete spec- trum from these generalized tadpoles (e.g., D. au- ratus) to aggressive, facultative carnivores (e.g., D. quinquevittatus) to obligate oophagous forms of the histrionicus SF)ecies group. Based on data reviewed here, we can make some predictions about the arboreal way of life for tad- poles. 1. Oophagy will be found, not in all arboreal tadpoles, but in ones with reduced numbers of denticle rows. 2. Tadpoles that are normally oophagous will feed on specialized, unfertilized, conspecific tro- phic eggs and not on fertilized eggs of either their own or other species. 3. Arboreal tadpoles with denticle formulas equal to or greater than 2/3 will be delriiivores. 4. Detritivorous arboreal tadpoles with normal body proportions and relatively short tails will be restricted to the larger aerial containers, such as tree holes. 5. Extremely elongate arboreal tadpoles will oc- cur in smaller aerial tanks, such as bromeliad leaf axils. 6. Arboreal tadpoles that normally live in hy- poxic tanks will have morphological and behav- ioral specializations to reduce oxygen loss to the water and to augment either aerial respiration or surface film respiration. Such features may in- clude: reduction of aquatic respiratory surfaces or reduced irrigation of such surfaces; low routine activity levels; vertical, head up swimming or rest- ing posture; large, patent, anteriorly directed glot- tis. Taxonomic Considerations Throughout this paper we have accepted O. brunneus and O. septentrionalis as congeners (Trueb & Tyler, 1974) for the sake of morpholog- ical consideration. We conclude with a comment on the reasonableness of this taxonomic associa- tion. In a classic paper on the concept of the genus, Inger (1958) argued that "each genus should rep- resent the same kind of entity: a distinct mode of life and a distinct evolutionary shift." It is quite apt that Inger should have used tadpoles to illus- trate his thesis. Following Inger's arguments, we consider it untenable that O. brunneus and O. sep- tentrionalis be included in the same genus, despite the gross morphological similarities of their adults. The life histories of these frogs, particularly the morphology and ecology of their larvae, are loo distinct— the evolutionary shift loo great— for Os- teopilus to be construed as a valid genus encom- passing both species. It may seem a simple mailer to return one or the other species to the genus Hyla, since both have been there before (cf Trueb & Tyler, 1974). However, this would not solve the problem, for Wassersug (1980), applying the same argument given above, has suggested that Hyla also does not form a single genus. Furthermore, we have not examined any specimens of the other species cur- rently assigned to Osteopilus, O. dominicensis. The syslemalics of Carribean hylids is, in fact, cur- rently under investigation by other workers using a variety of nonlarval characters (J. Bogart, R. Crombie, B. Hedges, pers. comm.). Because the purpose of this study is not system- atics, we resist making any formal taxonomic reas- signments. We leave the matter open, but predict that further quantitative studies on the genetic similarities of the species will confirm that brun- neus and septentrionalis should be in different gen- era. Summary We studied the morphology, ecology, and be- havior of the arboreal tadpole of the Jamaican hylid, Osteopilus brunneus. Included in our mor- phological descriptions is information on the cra- nial myology, abdominal viscera, internal oral anatonry, chondrocranium, lateral line system, and overall growth. Osteopilus brunneus tadpole mor- LANNOO ET AL.: LARVAL LIFE IN LEAVES 27 phology is compared to that of the currently rec- ognized congener, O. septentrionalis, which has a generaHzed pond larva. The ecological and be- havioral features that we examined in the labo- ratory included gut contents, normal resting pos- ture, aerial respiratory behavior, swimming performance, ability to survive subaerial expo- sure, and tolerance to aquatic hypoxia. In the field the habitat and conspecific associations of adults, tadpoles, and eggs were recorded. Our major results concerning O. brunneus lar- vae are as follows: 1 . In the field larvae always co-occur with eggs— found either as gut contents or nearby— even when adults are not breeding. 2. Their major food item is undeveloped, con- specific eggs. 3. The stomach is a major storage organ that can hold more than 1 80 eggs in a large tadpole. 4. Larval growth is not isometric; most of the elongation of the tadpole is because of dispropor- tionate tail elongation during early larval stages. 5. Osteopilus brunneus tadpoles are distin- guished from O. septentrionalis by {a) wider buccal floor and mouth; (b) reduced buccal floor and roof papillae; (c) enlarged anteriorly directed glottis; {d) smaller and shallower branchial baskets with re- duced gill filters; {e) reduced gill filaments; (f) no secretory ridges in the branchial food traps; {g) undivided dorsal velum; {h) enlarged orbitohyoi- deus and angularis musculature; (/) enlarged liver; (7) sacculate stomach; and (k) greatly reduced lat- eral line neuromasts. 6. The normal larval resting posture is with the snout pointing upward and the animal supported on the bottom by the terminal portion of the tail. 7. Larvae do not pump water through their buc- copharyngeal cavity; they open their mouths only to take in air or food. 8. They are light-shy, poor swimmers that (a) swim sporadically; {b) can only achieve moderate velocities— ca. 5 body lengths per second; (c) swim with excessive lateral movement; and {d) have low kinematic efficiency. 9. They are obligate air-breathers at all levels of dissolved oxygen that they are likely to expe- rience in the field. 10. They survive aquatic hypoxia and subaerial exposure longer than Rana and Xenopus tadpoles. 1 1 . They have a reduced cutaneous surface area for their length, perhaps to prevent oxygen loss to the water across these surfaces. 1 2. Contrary to the conclusion of Laessle (1961), the unusually elongate tail of these larvae is not used primarily to mix water nor is it specialized for aquatic respiration or aquatic locomotion. Most of the unique morphological features of O. brunneus larvae can be accounted for by the evolution of an early metamorphic onset in an otherwise generalized larva. Metamorphosis itself is not accelerated, but the metamorphic process seems to begin before all typical features of pond larvae have fully developed. Not all morphological features of these larvae, however, can be explained by this evolutionary mechanism (e.g., lateral line neuromasts, tail elongation). Oophagy in O. brunneus appears to involve spe- cialized trophic eggs as has been reported in Den- drobates pumilio, D. speciosus, and/), histrionicus. This conclusion is based on the seasonality of fer- tilized egg production and undeveloped eggs in- gested by larvae. Indirect evidence suggests that obligate oophagy in other arboreal tadpoles also involves specialized trophic eggs rather than sim- ply cannibalism. Based on what is known about the morphology and behavioral ecology of arboreal tadpoles in oth- er families and genera, we expand Orton's (1953) classification for tadpoles to include five distinct arboreal tadpole types: (1) elongate tadpoles with denticle rows <2/3; (2) stout tadpoles with den- ticle rows < 2/3; (3) elongate tadpoles with denticle rows >2/3; (4) stout tadpoles with denticle rows >2/3; and (5) Mertensophryne micranotis. Only types 1 and 2 are macrophagous or oophagous. Following the ideas of Inger (1958) on the con- cept of the genus, we conclude that Osteopilus brunneus is not a valid congener of O. septentrion- alis. Acknowledgments We thank James Bogart, Ron Crombie, James Dixon, Blair Hedges, Oswaldo Peixoto, William Pyburn, Barbara Zimmerman, and Helmut and Hike Zimmermann for discussing with us arboreal tadpole behavior and permitting us to cite their unpublished observations. Martha Crump made available Osteopilus septentrionalis tadpoles for this study. Neville Ellis provided field assistance and a cultural bridge. Calvin Cotterell of the Forest Service in Jamaica provided us with lodging and transportation. Malcolm Hendry and Steve Head at U.W.I, assisted with assorted logistical matters 28 FIELDIANA: ZOOLOGY in Jamaica. We are particularly grateful to Ronn Altig, Warren Burggren, Martha Crump, Robert Drewes, Roy McDiarmid, and our laboratory mates, Karin v. S. Hoff and V. Ann King, who critically reviewed drafts of this manuscript. The filming, digitizing, and analysis of swimming in O. brunneus would not have been possible without K. Hoffs expertise. Ann King additionally helped with darkroom work, illustration, and library work— all aspects of manuscript production. Fieldwork was supported by a National Geo- graphic Society Grant to D.S.T.; laboratory work was sponsored by a Natural Science and Engi- neering Research Grant (Canada) to R.J.W. We dedicate this paper to Robert F. Inger, ed- ucator and friend. His unrelenting commitment to excellence has been an inspiration to his students, and his student's students. Literature Cited JAdams, CD. 1972. Flowering Plants of Jamaica. The University Press, Glasgow, 848 pp. AuERCH, P., S. J. Gould. G. F. Oster, and D. B. Wake. 1979. Size and shape in ontogeny and phylogeny. Paleobiology, 5: 296-3 1 7. Alcala, a. C, and W. C. Brown. 1 982. Reproductive biology of some species of Philautus (Rhacophoridae) and other Philippine anurans. Philippine Journal of Biology, 11: 203-226. Alexander, R. M. 1968. Animal Mechanics. Sidgwick and Jackson, London, 346 pp. Altig. R., and J. P. Kelly. 1974. Indices of feeding in anuran tadpoles as indicated by gut characteristics. Herpetologica, 30: 200-203. Asprey, G. F.. and R. G. Robbins. 1953. The vege- tation of Jamaica. Ecological Monographs, 23: 359- 412. Barrington, E. J. W. 1 946. The delayed development of the stomach in the frog (Rana temporaria) and the toad (Bufo hufo). Proceedings of the Zoological Society of London, 116: 1-21. i Berry, P. Y. 1972. The Amphibian Fauna of Penin- sular Malaysia. Tropical Press, Kuala Lumpur, Ma- laysia, 1 30 pp. I Blommers-Schlosser, R. M. A. 1979. Biosystematics of the Malagasy frogs. I. Mantellinae (Ranidae). Beau- fortia, 29(352): 1-77. BoKERMANN, W. C. A. 1966. O Genero Phyllodytes Wagler, 1830 (Anura, Hylidae). Anais da Academia Brasileira de Ciencias, 38: 335-344. Bonner, J. T., ed. 1982. Evolution and Development. Dalhem Conference Report. Springer- Verlag, Berlin, 357 pp. 1 BouLENGER, G. A. 1903. Report on the batrachians and reptiles, pp. 131-171. In Annandale, N., and H. C. Robinson, eds.. Fasciculi Malayensis, vol. I. Long- mans, Green & Co., London. Brown, W. C, and A. C. Alcala. 1983. Modes of reproduction of Philippine anurans, pp. 416-428. In Rhodin. A., and K. Miyata, eds.. Advances in Her- petology and Evolutionary Biology. Museum of Com- parative Zoology, Cambridge, Mass. BuRGGREN. W., AND A. MwALUKOMA. 1983. Respi- ration during chronic hypoxia and hyperoxia in larval and adult bullfrogs. I. Morphological responses of lungs, gills, and skin. Journal of Experimental Biol- ogy, 105: 191-203. Burggren, W. W., AND N. H. West. 1983. Changing respiratory importance of gills, lungs and skin during metamorphosis in the bullfrog, Rana catesheiana. Respiratory Physiology, 47: 151-164. Channing, A. 1978. A new bufonid genus (Amphibia: Anura) from Rhodesia. Herpetologica, 34: 394-397. DE Jongh, H. J. 1968. Functional morphology of the jaw apparatus of larval and metamorphosing Rana temporaria L. Netherlands Journal of Zoology, 18: 1-103. Dixon, J. R., and C Rivero-Blanco. 1985. A new dendrobatid frog (Colostethus) from Venezuela with notes on its natural history and that of related species. Journal of Herpctology, 19: 177-184. Dring, J. C M. 1979. Amphibians and reptiles from northern Trengganu, Malaysia, with descriptions of two new geckos: Cnemaspis and Cyrtodactylus. Bul- letin of the British Museum of Natural History (Zool.), 34: 181-241. DuELLMAN, W. E. 1970. The hylid frogs of Middle America. Monograph of the Museum of Natural His- tory, The University of Kansas, 1: 1-753. DuELLMAN. W. E.. AND A. ScHWARTZ. 1958. Amphib- ians and reptiles of Southern Florida. Bulletin of the Florida State Museum. 3: 181-324. Dunn, E. R. 1926. The frogs of Jamaica. Proceedings of the Boston Society of Natural History, 38: III- 130. . 1937. The amphibian and reptilian fauna of bromcliads in Costa Rica and Panama. Cop>eia, 1937: 163-167. Feder, M. E. 1984. Consequences of aerial respiration for amphibian larvae, pp. 71-86. In Seymour, R. S., ed.. Respiration and Metabolism of Embryonic Ver- tebrates. Dr. W. Junk Publishers, Dordrecht, Neth- erlands. Feder, M. E., and W. W. Burggren. 1 984. Cutaneous gas exchange in vertebrates: Design, patterns, control and implications. Biological Reviews, 60: 1-45. Feder, M. E., and C. M. Moran. 1 985. Effect of water depth on costs of aerial respiration and its alternatives in tadpoles of Rana pipiens. Canadian Journal of Zo- ology, 63: 643-648. Feder, M. E., and R. Wassersug. 1 984. Aerial versus aquatic oxygen consumption in larvae of the clawed frog, Xenopus laevis. Journal of Experimental Biol- ogy, 108:231-245. Gans, C 1 975. The biological roles of taxonomic char- acteristics utilized in amphisbaenian classification. British Journal of Herpetology, 5: 61 1-615. LANNOO ET AL.: LARVAL LIFE IN LEAVES 29 GosNER, K. L. 1960. A simplified table for staging anuran embryos and larvae with notes on identifica- tion. Herpetologica, 16: 183-190. Grandison, a. G. C. 1 980. Aspects of breeding mor- phology in Mertensophryne micranotis (Anura: Bu- fonidae): Secondary sexual characters, eggs and tad- poles. Bulletin of the British Museum of Natural History (Zool.), 39: 299-304. Grandison, A. G. C, AND S. Ashe. 1983. The distri- bution, behavioural ecology and breeding strategy of the pygmy toad, Mertensophryne micranotis (Lov.). Bulletin of the British Museum of Natural History (Zool.), 45: 85-93. Harris, G. G., AND D. C. Milne. 1966. Input-output characteristics of the lateral line sense organs oi Xen- opus laevis. Journal of Acoustical Society of America, 40: 32-42. Heyer, W. R. 1973. Ecological interactions of frog larvae at a seasonal tropical location in Thailand. Jour- nal of Herpetology, 7: 337-361. HoFF, K., AND R. Wassersug. 1986. The kinematics of swimming in larvae of the clawed frog, Xenopus laevis. Journal of Experimental Biology, 122: 1-12. Inger, R. F. 1956. Some amphibians from the low- lands of North Borneo. Fieldiana: Zool., 34: 389-424. . 1958. Comments on the definition of genera. Evolution, 12: 370-384. . 1966. The systematics of zoogeography of the Amphibia of Borneo. Fieldiana: Zool., 52: 1-402. In press. Diets of tadpoles living in a Bomean rain forest. Alytes. Inger, R. F., H. B. Shaffer, M. Koshy, and R. Bakde. 1985. A report on a collection of amphibians and reptiles from Ponmudi, Kerala, South India. Journal of the Bombay Natural History Society, 81: 406-427, 551-570. Jones, D. A. 1967. Green pigmentation in Neotropical frogs. Doctoral dissertation. University of Horida, Gainesville, 1 54 pp. JuNGFER, K.-H. 1985. Beitrag zur Kenntnis von Den- drobates speciosus O. Schmidt, 1857 (Salientia: Den- drobatidae). Salamandra, 21: 263-280. Kenny, J. S. 1969. The Amphibia of Trinidad. Studies on the Fauna of Curasao and other Caribbean Islands No. 108, 29: 1-79. Kramer, D. L. 1983. The evolutionary ecology of res- piratory mode in fishes: An analysis based on the costs of breathing. Environmental Biology of Fishes, 9: 145-158. Laessle, a. M. 1961. A micro-limnological study of Jamaican bromeliads. Ecology, 42: 499-5 1 7. Lannoo, M. J. 1985. Neuromast topography in Am- bystoma larvae. Copeia, 1985: 535-539. LiMBAUGH, B. A., AND E. P. VoLPE. 1957. Early de- velopment of the Gulf Coast toad, Bufo valliceps Wieg- mann. American Museum Novitates, 1842: 1-32. LuTZ, A., AND B. LuTZ. 1939. New Hylidae from Bra- zil. Anais da Academia Brasileira de Ciencias, 11: 67- 89. LuTZ, B. 1973. Brazilian Species of Hyla. University of Texas Press, Austin, 265 pp. Marian, M. P., K. Sampath, A. R. C. Nirmala, and T. J. Pandian. 1980. Behavioural response of /?a/ia cyanophylictis tadpole exposed to changes in dissolved oxygen concentration. Physiology and Behavior, 25: 35-38. McDiarmid, R. W., and M. S. Foster. 1975. Unusual sites for two Neotropical tadpoles. Journal of Herpe- tology, 9: 264-265. Myers, C. W., J. W. Daly, and V. Martinez. 1984. An arboreal poison frog (Dendrobates) from Western Panama. American Museum Novitates, 2783: 1-20. Noble, G. K. 1927. The value of life-history data in the study of the evolution of the Amphibia. Annals of the New York Academy of Science, 30: 31-128. . 1 929. The adaptive modifications of the ar- boreal tadpoles of Hoplophryne and the torrent tad- poles of Staurois. Bulletin of the American Museum of Natural History, 83: 291-336. -. 1931. The Biology of the Amphibia. McGraw- Hill, New York, 577 pp. Orton, G. L. 1944. Studies on the systematic and phylogenetic significance of certain larval characters in the Amphibia Salientia. Doctoral dissertation. Uni- versity of Michigan, Ann Arbor, 254 pp. . 1953. The systematics of vertebrate larvae. Systematic Zoology, 2: 63-75. Parker, H. W. 1934. A Monograph of the Frogs of the Family Microhylidae. British Museum of Natural History, London, 208 pp. Peixoto, O. L. 1981. Notas sobre a girino de Cros- sodactylodes pintoi Cochran (Amphibia, Anura, Lep- todactylidae). Revista Brasileira Biologia, 41: 339-341. . 1983. Jvjo ntvj %Tpec\ts of Crossodactylodes oi Santa Tereza, state of Espirito Santo, Brazil (Am- phibia, Anura, Leptodactylidae). Revista Brasileira Biologia, 42: 619-626. Perret, J.-L. 1962. La biologic d'Acanthixalus spi- nosus (Amphibia Salientia). Recherches Etudes Ca- merounaises, 1: 90-101. Razarihelisoa, M. 1974. Contribution a I'etude des batraciens de Madagascar. Ecologie et developpement larvaire de Gephyromantis methueni. Angel, batracien a biotope vegetal sur les Pandanus. Bulletin Academic Malgache, 51: 113-128. Robinson, D. C. 1977. Herpetofauna bromelicola Cos- tarricense y Renacuajos de Hyla picadoi Dunn. His- toria Natural de Costa Rica, 1: 31-42. Salthe, S. N., and J. S. Mecham. 1 974. Reproductive and courtship patterns, pp. 309-521. In Lofts, B., ed., Physiology of the Amphibia, vol. II. Academic Press, New York. Satel, S., and R. Wassersug. 1981. On the relative size of the buccal floor depressor and elevator muscles in tadpoles. Copeia, 1981: 129-137. Savage, R. M. 1952. Ecological, physiological and an- atomical observations on some sp)ecies of anuran tad- poles. Proceedings of the Zoological Society of Lon- don, 122:467-514. Schwartz, A., and D. C. Fowler. 1973. The anura of Jamaica: A progress report. Studies of the Fauna of Cura9ao and other Caribbean Islands, 43: 50-142. Shreckenberg, M. G. 1956. The embryonic devel- 30 HELDIANA: ZOOLOGY > opment of the thyroid gland in the frog, Hvla brunnea. Growth, 20: 295-313. SiLVERSTONE, P. A. 1975. A revision of the poison- arrow frogs of the genus Dendrobates Wagler. Natural History Museum of Los Angeles County Science Bul- letin, 21: 1-55. 1976. A revision of the poison-arrow frogs of the genus Phyllobates Bibron in Sagra (family Den- drobatidae). Natural Historv' Museum of Los Angeles County Science Bulletin. 27: 1-53. Starrett, p. 1 960. Description of tadpoles of Middle American frogs. Miscellaneous Publications of the Museum of Zoology. University of Michigan, 110: 1- 37. Stuart. L. C. 1948. The amphibians and reptiles of Alta Verapaz, Guatemala. Miscellaneous Publications of the Museum of Zoologv, University of Michigan. 69: 1-109. Taylor. E. H. 1940. Two new anuran amphibians from Mexico. Proceedings of the U.S. National Mu- seum, 89: 43-47. . 1954. Frog-egg-eating tadpoles of .-Iwor/ifcaco- ronata (Stejneger) (Salientia, Hylidae). University of Kansas Science Bulletin, 36: 589-596. . 1962. The amphibian fauna of Thailand. Uni- versity of Kansas Science Bulletin, 43: 265-599. Trueb, L., and M. J. Tyler. 1974. Systematics and evolution of the Greater Antillean hylid frogs. Occa- sional Papers of the Museum of Natural History, The University of Kansas, 24: 1-60. Valerio, C. E. 1 97 1 . Ability of some tropical tadpoles to survive without water. Copeia, 1971: 364-365. VoGEL, S. 1 98 1 . Life in Moving Fluids. Willard Grant Press, Boston, 352 pp. VoLPE, E. P. 1957. The early development of Rana capita sevosa. Tulane Studies in Zoology, 5: 207-225. Wassersug, R. J. 1976. Oral morphology of anuran larvae: Terminology and general description. Occa- sional Papers of the Museum of Natural History, Uni- versity of Kansas, 48: 1-23. . 1 980. Internal oral features of larvae from eight anuran families: Functional, systematic, evolutionary and ecological considerations. Miscellaneous Publi- cations of the Museum of Natural History, University of Kansas, 68: 1-146. Wassersug, R. J., and W. E. Duellman. 1984. Oral structures and their development in egg-brooding hy- lid frog embryos and larvae: Evolutionary and eco- logical implications. Journal of Morphology, 182: 1- 37. Wassersug, R. J., and M. E. Feder. 1 983. The effects of oxygen concentration, body size and respiratory behaviors on the stamina of obligate aquatic (Bufo americanus) and facultative air breathing (Xenopus laevis and Rana berlandieri) anuran larvae. Journal of Experimental Biology, 105: 173-190. Wassersug, R. J., K. J. Frogner, and R. F. Inger. 1981. Adaptations for life in tree holes by rhaco- phorid tadpoles from Thailand. Journal of Herpetol- ogy. 15:41-52. Wassersug. R. J., and W. R. Heyer. 1983. Morpho- logical correlates of sub-aerial existence in leptoidac- tylid tadpoles associated with flowing water. Canadian Journal of Zoology, 61: 761-769. Wassersug. R.. and K. Hoff. 1979. A comparative study of the buccal pumping mechanism of tadpoles. Biological Journal of Linnean Society, 12: 225-259. . 1982. Developmental changes in the orienta- tion of the anuran jaw suspension. Evolutionary Bi- ology, 15: 223-246. 1985. The kinematics of swimming in anuran larvae. Journal of Experimental Biology, 119: 1-30. Wassersug, R. J., and K. Rosenberg. 1979. Surface anatomy of branchial food traps of tadpoles: A com- parative study. Journal of Morphology, 159: 393-423. Wassersug, R.J. .AND E. A. Seibert. 1975. Behavioral responses of amphibian larvae to variation in dis- solved oxygen. Cop>eia, 1975: 86-103. Weygoldt, p. 1980. Complex brood care and repro- ductive behavior in captive poison-arrow frogs, Den- drobates puniilio O. Schmidt. Behavioral Ecology and Sociobiology, 7: 329-332. . 1 987. Evolution of parental care in dart poison frogs (Amphibia: Anura: Dendrobatidae). Zeitschrift fiirzoologischeSystemikundEvolutionsforschung,25: 51-67. WiLCZYNSKA, B. 1981. The Structure of the alimentary canal and the dimensions of the mucosa in ontogenetic development of some Anura. Acta Biologica Craco- viensia (Zool.), 23: 1 3-46. ZiMMERMANN, E.. AND H. ZlMMERMANN. 1985. Brut- pflegestrategien bei Pfeilgiftfroschen (Dentrobatidae). Vcrhandlungen der Deutschen Zoologischen Gesell- schaft. 78: 220. ZlMMERMANN, H. 1974. Die Aufzucht dcs Goldbaum- steigers Dendrobates auratus. Aquarien Magazin, 8: 526-531. . 1982. Durch Nachzucht erhalten: Blatt- steigerfrosche Phyllobates vittatus und P. lugubris. Aquarien Magazin, 16: 109-1 12. ZlMMERMANN, H.. AND E. ZlMMERMANN. 1980. Durch Nachzucht erhalten: Der Baumsteiger Dendrobates leucomelas. Aquarien Magazin. 14: 21 1-217. . 1981. Social verhalten, Fortpflanzungsverhal- ten und Zucht der Farberfrosche Dendrobates histri- onicus und D. lehmanni sowie einiger anderer Den- drobatiden. Zeitschrift des Kolner Zoo, 24: 83-99. . 1984. Durch nachzucht erhalten: Baumstei- gerfrosche Dendrobates quinquevitlatus und D. retic- ulars. Aquarien Magazin, 18: 35-41. LANNOO ET AL.: LARVAL LIFE IN LEAVES 31 Field Museum of Natural History Roosevelt Road at Lake Shore Drive Chicago, Illinois 60605-2496 Telephone: (312) 922-9410