Female offspring communication in a Taiwanese tree frog, Chirixalus eiffingeri (Anura: Rhacophoridae)

Transcription

1 ANIMAL BEHAVIOUR, 2002, 64, doi: /anbe , available online at Female offspring communication in a Taiwanese tree frog, Chirixalus eiffingeri (Anura: Rhacophoridae) YEONG-CHOY KAM & HUI-WEN YANG Department of Biology, National Changhua University of Education ((Received 12 October 2001; initial acceptance 3 January 2002; final acceptance 14 May 2002; MS. number: 7096) ) We assessed the roles of visual and olfactory cues in female tadpole communication in Chirixalus eiffingeri. The mean cumulative time that at least one tadpole was active or begged for food was significantly longer when a female C. eiffingeri was present than when a plastic frog was introduced and when no frog was present. Tadpoles did not respond visually to a female frog physically separated from them by transparent Plexiglas. However, tadpoles were more active in water conditioned by female frogs than in unconditioned water. Tadpole activity was further elevated by water conditioned by a female frog and tadpoles. Tadpoles were more active in water conditioned by male frogs than in unconditioned water, but water conditioned by a male frog and tadpoles did not further elevate tadpole activity. Thus, water conditioned by adults of either sex contains substances that increase tadpole activity, but only females show a synergistic effect with conditioning by tadpoles Published by Elsevier Science Ltd on behalf of The Association for the Study of Animal Behaviour Oviposition by frogs in phytotelmata, unique aquatic microhabitats, is probably an evolutionary adaptation for avoiding the predators that inhabit larger aquatic habitats (Duellman & Trueb 1986; Lannoo et al. 1987; Summers 1990; Caldwell 1993; Jungfer 1996; Kam et al. 1996). Maternal provisioning of eggs to feed arboreal tadpoles probably evolved in response to food scarcity in phytotelmata (Wassersug et al. 1981). Laboratory and field studies have revealed that egg provisioning may be biparental or uniparental, and that female tadpole communication is particularly important in uniparental care. In biparental care, the female lays eggs to feed the tadpoles in the presence of a male frog. During provisioning, male and female frogs return to the hole, engage in courtship behaviour, and lay fertilized or unfertilized eggs in the pool (Caldwell & deoliveira 1999). Fertilized eggs that are not consumed by the tadpoles will later hatch and become the next cohort of tadpoles. Biparental care occurs in Osteopilus brunneus (Thompson 1996), Osteocephalus oophagous (Jungfer & Weygoldt 1999), and Dendrobates vanzolinii (Caldwell & deoliveira 1999). In uniparental care, the female frog lays unfertilized eggs to feed the tadpoles in the absence of male frogs. Interactions between a female frog and tadpoles seem to signal or induce the female to lay trophic eggs Correspondence and present address: Y.-C. Kam, Department of Biology, Tunghai University, Taichung 407, Taiwan, R.O.C. ( biyckam@mail.thu.edu.tw). H.-W. Yang is at the Department of Biology, National Changhua University of Education, Changhua 50058, Taiwan, R.O.C. for the tadpoles. Chirixalus eiffingeri tadpoles, which are obligatorily oophagous, are fed intermittently by females that lay unfertilized trophic eggs (Kam et al. 1996). When a female frog returns to the nest, tadpoles immediately aggregate around her. Each tadpole stiffens its tail and begins vibrating vigorously, while nipping at the skin around her cloaca and thighs (Ueda 1986). The female begins to lay trophic eggs, a few at a time, although no males are present. As soon as the eggs are laid, the tadpoles bite them and suck out the yolk (Ueda 1986). Similar egg-begging behaviour has also been reported in a dendrobatid frog (Dendrobates pumilio) and two hylids (Anotheca spinosa and O. brunneus) from the New World tropics (Weygoldt 1980; Jungfer 1996; Thompson 1996). To date, studies of female tadpole communication have been mostly observational, and the communication between female frogs and tadpoles has not been studied experimentally. Jungfer (1996) studied the brooding behaviour of A. spinosa and showed that tactile stimulation by the tadpoles was essential for inducing female frogs to lay trophic eggs. Anotheca spinosa and C. eiffingeri tadpoles nip the female s skin throughout of a feeding event. This behaviour is fastest and most vigorous moments before eggs are deposited in the pool (Jungfer 1996; Y.-C. Kam, unpublished data). Why do tadpoles continuously nip the skin of female frogs? If tactile stimulation of the female frog by tadpoles is necessary to induce her to lay eggs, she might communicate with the tadpoles to encourage their nipping /02/$35.00/ Published by Elsevier Science Ltd on behalf of The Association for the Study of Animal Behaviour

2 882 ANIMAL BEHAVIOUR, 64, 6 We investigated experimentally female tadpole communication in C. eiffingeri. We assessed in particular the possible role of visual and olfactory cues in this communication and the properties of the chemical cues. Chirixalus eiffingeri is a small frog (snout vent length ca mm), endemic to Taiwan and two adjacent small islands, Iriomote and Ishigaki (Kuramoto 1973; Ueda 1986). During the breeding season (February August), male frogs call from bamboo stumps. Frequently, more than one male frog occupies a bamboo stump, and they compete for females. Female frogs deposit fertilized eggs above the waterline on the inner walls of tree holes or bamboo stumps (Kuramoto 1973; Kam et al. 1998a). Upon hatching, tadpoles drop into the pool of water where they grow and develop until metamorphosis. Male frogs moisten the eggs during the embryonic stage, but leave the stumps after the embryos have hatched. Female frogs visit and feed tadpoles at night at intervals of about 8 days (Kam et al. 2000). The length of the larval period, from hatching to metamorphosis, is days (Kam et al. 1998b). Tadpoles are not cannibalistic, but they sometimes scavenge the remains of dead siblings. The effects of scavenging on the growth and development of tadpoles are negligible (Kam et al. 1996, 1998b, 2000). Study Animals GENERAL METHODS In April 1999, we collected 10 pairs of C. eiffingeri in amplexus from the bamboo forests of Chitou, Nantou County, and transported them to our laboratory in Changhua. Each pair of frogs was maintained in a glasssided terrarium (42 27 cm and 30 cm high) containing a layer of moist soil, plants, and a bamboo stump half filled with water. The frogs were fed house flies ad libitum. The terraria were placed in an animal room at C and a 12:12 h photoperiod. All frogs successfully deposited fertilized eggs in the bamboo stumps in the terraria (between May and September). Female frogs provisioned eggs to feed tadpoles, as evidenced by the growth of tadpoles. We collected about 340 tadpoles from bamboo stumps in the forest. Only tadpoles at Gosner stages (Gosner 1960) were used. Each tadpole was used once in each experiment. Each female frog was used only once in each experiment, but owing to a shortage of female frogs during the study period, each female was used in more than one experiment. At the end of the experiments, all frogs were returned and released in the bamboo forest in Chitou. At the end of each experiment, we transported the tadpoles to Chitou and put them in bamboo stumps that already contained tadpoles. Because female C. eiffingeri feed both kin and nonkin (Kam et al. 2000), all tadpoles in the bamboo stumps were fed and grew normally, eventually reaching metamorphosis (Kam et al. 1998a; Chen et al. 2001). We fed tadpoles with chicken egg yolk once every 4 days until they were returned to the wild. An earlier study showed that chicken egg yolk is a good substitute for C. eiffingeri eggs. Furthermore, tadpoles that were fed once every 4 days grew as well as tadpoles that were fed by a female frog (Liang et al. 2002). The study was conducted in accordance with the legal requirements of the National Science Council. Statistics The data were analysed with nonparametric statistics (SAS Institute Inc. 1996). The Mann Whitney test or Kruskal Wallis test was used to evaluate tadpole responses when samples were independent, and the Wilcoxon test or Friedman test for matched samples. If, and only if, the overall effect was significant (Kruskal Wallis test or Friedman test), we then performed multiple comparisons (Mann Whitney test or Wilcoxon test, respectively) to determine which pairs of treatments were significantly different. For all variables, means SD are reported unless otherwise noted and two-tailed tests were used. Methods EXPERIMENT 1: PRESENCE OF FEMALE We assessed tadpole activity under three conditions: (1) when a female frog was present; (2) when a plastic frog of similar size was present; and (3) when a frog was not present. For each trial, we placed 10 tadpoles in a plastic container (9 9 cm and 7 cm high) in distilled water 1.5 cm deep. We placed a female or plastic frog in the container and lowered a styrofoam lid to 1 cm above the water to prevent the female escaping. After allowing 30 s for the animals to settle down, we videotaped the tadpoles through the bottom for 5 min. Using the same procedures, we also videotaped tadpoles in the container for 5 min when no frogs were added. Each group of 10 tadpoles was subjected to each treatment only once. The treatment order for each group of tadpoles was assigned randomly. We conducted eight trials for each treatment. We assessed tadpole responses by measuring two variables: (1) the cumulative time at least one tadpole was active; and (2) the cumulative time at least one tadpole begged for eggs over a 5-min period. Inactivity was defined as the absence of forward movement. A tadpole moving its tail slightly without causing forward body movement was considered inactive. A tadpole engaged in egg-begging behaviour stiffened its tail and vibrated vigorously, often nipping the skin of any female frog that was present. Egg begging was conspicuous and easy to differentiate from the undulating tail movements typical of tadpoles when females were absent. Results The mean cumulative time at least one tadpole was active varied significantly between treatments (Friedman test: χ 2 2=13.12, P<0.01; Fig. 1). Tadpoles were most active when a female frog was present and significantly more active than tadpoles in the container with a plastic frog or no frog (Wilcoxon: test: T=3.37, N=8, P=0.005 in both

3 KAM & YANG: ANURAN FEMALE OFFSPRING COMMUNICATION 883 Mean cumulative time (s) Female frog Plastic frog None Treatment Figure 1. The mean cumulative time at least one tadpole was active or begged for eggs in a 5-min period when a female frog, a plastic frog, or no frogs were present. Values are means±sd. N=8 replicates. cases). The activity of tadpoles in the treatment with a plastic frog and without a frog did not differ significantly (Wilcoxon test; T=0.42, N=8, P=0.72). The mean cumulative time at least one tadpole was begging also varied between treatments (Fig. 1), with tadpoles begging more when a female was present than when a plastic frog was present (Wilcoxon test: T=3.37, N=8, P=0.0009). Tadpoles did not beg in the no-frog treatment. In the treatment where the female frog was present, the mean cumulative time tadpoles were active was significantly correlated with the time tadpoles spent begging (Spearman rank correlation: r S =0.977, N=8, P=0.0001). In all eight trials with female frogs, tadpoles actively nipped the females. Some female frogs sat quietly as tadpoles nipped them, but others moved around. Some females appeared reluctant to be nipped by the tadpoles and used their hindlegs to push tadpoles away. These females tried to leave the water, but could not because of the lid. None of the female frogs laid unfertilized eggs in response to the begging of the tadpoles. EXPERIMENT 2: Active Begging VISUAL AND CHEMICAL CUES Methods Visual cues We placed a female frog in a small, transparent, Plexiglas container (9 9 cm and 7 cm high) and lowered a perforated styrofoam lid to 1 cm above the water to prevent her escaping. The container was half submerged and suspended in a larger, transparent, Plexiglas container (15 15 cm and 10 cm high) containing distilled water and 10 tadpoles. This set-up allowed tadpoles to see the female from the side and from below. After allowing 30 s for the animals to settle, we videotaped the tadpoles through the bottom of the large container for 5 min. The control treatment used the same set-up but there was no female frog in the small container. Each group of 10 tadpoles was subjected to each treatment once. For each group of tadpoles, the treatment order was assigned randomly. We conducted eight trials of each treatment. We counted the cumulative time at least one tadpole was active during a 5-min period using the criteria described in experiment 1. Olfactory cues To assess the olfactory response of tadpoles, we used distilled water conditioned by female frogs (treatment group) and distilled water (control group). We used distilled water because it was more comparable to natural conditions in that the water in nests (tree holes or bamboo stumps) is derived primarily from rain. Prior to the experiment, one female frog was immersed for 4 h in each of 10 beakers containing 50 ml of distilled water. The conditioned water from each beaker was filtered and then the water from the 10 beakers was pooled. In each trial, we poured 50 ml of conditioned water into plastic cup A and 50 ml of distilled water into plastic cup B. A tadpole was placed in each cup, and videotaped for 5 min, after allowing 30 s for settling. After videotaping, we placed the tadpoles in separate beakers containing distilled water for a 1 min rest. Then, the tadpole that had been in cup A was put in cup B and the tadpole from cup B was placed in cup A. After allowing 30 s for settling, we videotaped both tadpoles for 5 min. This design minimized bias caused by individual differences in activity. However, by doing so, we subjected the second tadpole to distilled water that was conditioned briefly by the first tadpole. Similarly, the second tadpole was subjected to female-conditioned water that was conditioned briefly by the first tadpole. We ran statistical tests and found that the presence of the first tadpole in distilled water or female-conditioned water had no effect on the activity level of the second tadpole in the subsequent experiments (Mann Whitney test: U=137, N 1 =N=15, NS; U=127, N 1 =N 2 =15, NS, respectively). We therefore averaged the active time of two tadpoles in each condition to give a measure of tadpole response. There were 15 measurements (trials) of tadpole response of each treatment. We assessed tadpole response by measuring the cumulative time each tadpole was active. Results Tadpoles did not respond to the female frog in the visual experiment. They moved little whether a female frog was present ( s, N=8) or absent ( s, N=8). The mean cumulative active time did not differ significantly between treatments (Wilcoxon test: T= 0.11, N=8, P=0.958). During videotaping, some female frogs moved around and stretched. The tadpoles did not respond to the female s movements and remained immobile most of the time.

4 884 ANIMAL BEHAVIOUR, 64, 6 Table 1. The activity levels (time spent active, s) of tadpoles in water conditioned by adult frogs (females or males) alone, adult frogs (females or males) and tadpoles, and tadpoles alone Adults Adult frogs alone Females 22.3±7.5 (7.0±2.7) Males 18.2±5.9 (6.0±5.9) In the olfactory experiment, tadpoles in water conditioned by female frogs moved significantly more than those in distilled water (Wilcoxon test: T=4.62, N=15, P=0.0001; Table 1). EXPERIMENT 3: Adult frogs and tadpoles 50.4±24.8 (8.0±4.5) 21.7±17.1 (9.0±6.1) CHEMICAL CUES Tadpoles alone 6.2±5.0 (4.0±4.2) The values in parentheses are the activity levels of tadpoles in unconditioned, distilled water. Values are reported as mean±sd, N=15 replicates. Methods Sex-specific chemical cues In this experiment, we evaluated tadpole responses to water conditioned by male and female frogs. We used the data from experiment 2 for the tadpole response to water conditioned by female frogs and the methods from experiment 2 to measure the tadpole response to water conditioned by male frogs. There were 15 replicates of each treatment. We measured the tadpole response by calculating the mean cumulative time tadpoles were active. We first compared, with the Wilcoxon test, the tadpole responses in the two treatments (distilled water versus conditioned water). Then, with the Mann Whitney test we compared the tadpole responses in water conditioned by female frogs versus water conditioned by male frogs. However, because the tadpole response in distilled water varied slightly each time, we adjusted it (called adjusted tadpole response hereafter) by subtracting tadpole response time in distilled water from response time in conditioned water. Female frog versus tadpoles We assessed tadpole responses to water conditioned by a female frog, a female frog and tadpoles, and tadpoles only. We used data from experiment 2 for the tadpole response to water conditioned by female frogs. To prepare the water conditioned by a female and tadpoles, we placed a female frog and 10 tadpoles for 4 h in each of 10 beakers containing 50 ml of distilled water. The frogs and tadpoles were removed and the conditioned water was filtered and then pooled. The same procedures were used to prepare the water conditioned by tadpoles only. We used the methods from experiment 2 to measure the tadpole response to water conditioned by female frogs and tadpoles and tadpoles alone. There were 15 replicates of each treatment. We measured the tadpole response by calculating the mean cumulative time tadpoles were active. We first compared with the Wilcoxon test the tadpole responses in two treatments (distilled water versus conditioned water) for each of the three experiments. Then, with the Kruskal Wallis test we compared the adjusted tadpole responses in water conditioned by female frogs, female frogs and tadpoles, and tadpoles alone. Multiple comparisons (Mann Whitney test) were performed if, and only if, the overall effect was significant. Male frog versus tadpoles We assessed tadpole response to water conditioned by male frogs only and by male frogs and tadpoles. We used data from experiment 3 as a measure of tadpole response to water conditioned by male frogs and distilled water. To prepare water conditioned by a male and tadpoles, we put 50 ml of distilled water in each of 10 beakers and then placed 10 tadpoles and a male frog in each beaker. After 4 h, the frog and tadpoles were removed, the conditioned distilled water from each beaker was filtered, and then pooled. We assessed tadpole response with the same methods used in experiment 2. There were 15 replicates in each treatment. We measured the tadpole response by calculating the mean cumulative time tadpoles were active. We first compared with the Wilcoxon test the tadpole responses in two treatments (distilled water versus conditioned water) for each experiment. Then, with the Mann Whitney test we compared the adjusted tadpole responses in water conditioned by male frogs versus water conditioned by male frogs and tadpoles. Results Sex-specific chemical cues Tadpoles in water conditioned by male frogs moved significantly more than those in distilled water (Wilcoxon test: T=3.96, N=15, P=0.0001; Table 1). There was no significant difference in the adjusted tadpole responses to water conditioned by male frogs or female frogs (data from experiment 2; Mann Whitney test: U=139, N 1 =N 2 =15, NS; Table 1). Female frog versus tadpoles Tadpoles in the water conditioned by a female frog and tadpoles moved significantly more than those in distilled water (Wilcoxon test: T=4.59, N=15, P=0.0001; Table 1). The activity level of tadpoles in water conditioned by tadpoles only did not differ significantly from that in distilled water (Wilcoxon test: T=1.22, N=15, P=0.236; Table 1). The adjusted tadpole responses to water conditioned by a female frog only (data from experiment 2), a female and tadpoles, or tadpoles only, were significantly different (Kruskal Wallis Test: χ 2 2=30.609, P=0.0001). The adjusted tadpole response in water conditioned by a female frog only or a female and tadpoles was significantly longer than in water conditioned by tadpoles alone (Mann Whitney test: U=214, N 1 =N 2 =15, P<0.001; U=222, N 1 =N 2 =15, P<0.001, respectively). Furthermore, the adjusted tadpole response in water conditioned by a female and tadpoles was significantly longer than in

5 KAM & YANG: ANURAN FEMALE OFFSPRING COMMUNICATION 885 water conditioned by a female alone (Mann Whitney test: U=188, N 1 =N 2, P<0.001). Male frog versus tadpoles Tadpoles in water conditioned by a male frog and tadpoles moved significantly more than those in the distilled water (Wilcoxon test: T=2.55, N=15 P=0.011; Table 1). There was no significant difference in the adjusted tadpole responses to water conditioned by a male frog (data from experiment 3) or by a male frog and tadpoles (Mann Whitney test: U=130, N 1 =N 2 = 15, NS). DISCUSSION In C. eiffingeri and other anurans with uniparental care, communication between females and their arboreal tadpoles involves complex behaviours mediated by tactile and chemical cues. Begging and provisioning behaviours are clearly the bases for the evolution of this communication system (Wilson 1980; Hoff et al. 1999). The most conspicuous form of communication between C. eiffingeri tadpoles and female frogs is egg-begging behaviour. This behaviour has been described in detail for C. eiffingeri (Ueda 1986), A. spinosa (Jungfer 1996) and D. pumilio (Weygoldt 1980; Brust 1993). Chirixalus eiffingeri tadpoles respond vigorously to conspecific males but not to female frogs of a different species (Ueda 1986); however, male frogs immediately move away when tadpoles nip them. In our experiments, we found that some females responded negatively to being nipped by tadpoles and quickly moved away. Some tadpoles clung to the skin of female frogs as they climbed out of the water (Y.-C. Kam, personal observation). We speculate that the tadpoles nipping is discomforting or irritating to the female frogs. We also suggest that the physiological readiness of a female frog strongly affects her response to tactile stimulation from tadpoles. A female frog may be more tolerant of tadpole nipping when she has mature eggs than when she does not. In experiment I, female frogs did not lay eggs during the 5 min they were confined with tadpoles, probably because the interaction between tadpoles and female frogs was too short, or female frogs were simply not ready, or willing, to lay eggs. Chirixalus eiffingeri tadpoles did not respond visually to a female frog, regardless of whether she was moving. This is consistent with the results of a preliminary laboratory experiment on interactions between a C. eiffingeri female and her offspring. We reared a pair of C. eiffingeri in a terrarium where they deposited fertilized eggs in a bamboo stump containing a pool of water. For a week, we videotaped the bamboo stump from 1800 to 0600 hours. The female C. eiffingeri slowly crawled up the outside of the bamboo stump, stopping for a while, then continuing. During her ascent, tadpoles in the water pool remained immobile. As soon as the female reached the rim of the stump, she jumped into the pool. The tadpoles immediately became extremely excited and started begging. In contrast, visual cues are important components of female tadpole communication in D. pumilio (Weygoldt 1980; Brust 1993). The tadpoles show a remarkable behaviour whenever an adult frog attempts to enter their leaf axils. This behaviour often starts before a frog has entered the leaf axil, triggered either by visual cues of the approaching frog or by vibrations caused by its movements (Weygoldt 1980). Instead of swimming with the usual undulating tail movements, the tadpoles stiffen their tails and rapidly vibrate them, producing conspicuous circular movements just beneath the water surface. A frog motivated to bathe is deterred by these movements, often leaving before it has touched the water and moving on to an empty axil for bathing (Weygoldt 1980). Dendrobates pumilio is diurnal, whereas C. eiffingeri is nocturnal, which probably explains the greater importance of visual cues to female tadpole communication in D. pumilio. Tadpoles placed in water conditioned by female frogs (experiment 2) increased their activity but did not beg. Thus, female frogs seem to give off water-soluble substances that stimulate tadpole activity. However, a female must be physically present to induce begging. Tadpole activity was stimulated even more by the water conditioned by a female frog and tadpoles, suggesting the female and tadpoles interact in a synergistic manner. The tadpole activity in water conditioned by a female frog and tadpoles was far greater than the sum of the tadpole activity in water conditioned by a female frog alone and tadpoles alone (see experiment 2). Although water conditioned by male frogs also elevated tadpole activity, water conditioned by a male frog and tadpoles did not increase tadpole activity further. Ueda (1986) reported that C. eiffingeri tadpoles vigorously nip the skin of male frogs, but the males move away as soon as they are nipped. Thus, certain substances attractive to tadpoles are given off by both sexes of C. eiffingeri. However, only female frogs tolerate being nipped by tadpoles. If female frogs are willing to be nipped by tadpoles, tadpole activity or movement may become faster and more vigorous as the encounter progresses; in some cases this communication triggers egg laying (Y.-C. Kam, personal observation). We hypothesize that the tactile stimulation from the tadpoles mimics the tactile stimulation of male frogs during amplexus, which induces the female frog to lay eggs. Chemical cues appear to play an important role in female tadpole communication. In amphibians, chemical communication is a component of kin recognition by anuran larvae (Waldman 1985, 1991; Blaustein & Walls 1995), predator avoidance (Manteifel 1995; Lefcort 1996; Summey & Mathis 1996), and reproductive behaviour (Forester 1986; Kikuyama et al. 1995). However, chemical communication between female frogs and tadpoles is novel. We do not know the identity of the chemical compounds involved in female tadpole interactions. To understand fully the scope and importance of chemical communication between female frogs and tadpoles, further experimental studies and identification of the active compounds are necessary. In addition, comparative studies of other species whose tadpoles show eggbegging behaviour, such as A. spinosa and D. pumilio, will be crucial.

6 886 ANIMAL BEHAVIOUR, 64, 6 Acknowledgments This study was supported by a National Science Council Grant (NSC B ) to Y.-C.K. We thank the staff of the Experimental Forest of the National Taiwan University at Chitou for providing accommodation and permitting us to collect specimens in the experimental forest. Comments and suggestions on the manuscript by A. F. Warneke are appreciated. References Blaustein, A. R. & Walls, S. C Aggregation and kin recognition. In Amphibian Biology. Vol. 2, Social Behaviour (Ed. By H. Heatwole & B. K. Sullivan), pp Chipping Norton, New South Wales: Surrey Beatty. Brust, D. G Maternal brood care by Dendrobates pumilio: a frog that feeds its young. Journal of Herpetology, 27, Caldwell, J. P Brazil nut fruit capsules as phytotelmata: interactions among anuran and insect larvae. Canadian Journal of Zoology, 71, Caldwell, J. P. & deoliveira, V. L Determinants of biparental care in the spotted poison frog, Dendrobates vanzolinii (Anura: Dendrobatidae). Copeia, 1999, Chen, Y.-H., Su, Y.-J., Lin, Y.-S. & Kam, Y.-C Interand intraclutch competition among oophagous tadpoles of the Taiwanese tree frog, Chirixalus eiffingeri (Anura: Rhacophoridae). Herpetologica, 57, Duellman, W. E. & Trueb, L Biology of Amphibians. New York: McGraw-Hill. Forester, D The recognition and use of chemical signals by a nesting salamander. In: Chemical Signals in Vertebrates 4 (Ed.byD. Duvall, D. Muller-Schwarze & R. M. Silverstein), pp New York: Plenum. Gosner, K. L A simplified table for staging anuran embryos and larvae with notes on identification. Herpetologica, 18, Hoff, K. V., Blaustein, A. R., McDiarmid, R. W. & Altig, R Behaviour. In: Tadpoles (Ed. by R. W. McDiarmid & R. Altig), pp Chicago: University of Chicago Press. Jungfer, K.-H Reproduction and parental care of the coronated treefrog, Anotheca spinosa (Steindachner, 1864) (Anura: Hylidae). Herpetologica, 52, Jungfer, K.-H. & Weygoldt, P Biparental care in the tadpolefeeding Amazonian treefrog Osteocephalus oophagus. Amphibia- Reptilia, 20, Kam, Y.-C., Chuang, Z.-S. & Yen, C.-F Reproduction, oviposition-site selection and larval oophagy of an aboreal nester, Chirixalus eiffingeri (Rhacophoridae), from Taiwan. Journal of Herpetology, 30, Kam, Y.-C., Lin, C.-F., Lin, Y.-S. & Tsai, Y.-F. 1998a. Density effects of oophagous tadpoles of Chirixalus eiffingeri (Anura: Rhacophoridae): importance of maternal brood care. Herpetologica, 54, Kam, Y.-C., Yen, C.-F. & Hsu, J.-L. 1998b. Water balance, growth, development, and survival of arboreal frog eggs (Chirixalus eiffingeri, Rhacophoridae): importance of egg distribution in bamboo stumps. Physiological Zoology, 71, Kam, Y.-C., Chen, Y.-H., Chen, T.-C. & Tsai, I.-R Maternal brood care of an arboreal breeder, Chirixalus eiffingeri (Anura: Rhacophoridae) from Taiwan. Behaviour, 37, Kikuyama, S., Toyoda, F., Ohmiya, Y., Matsuda, K., Tanaka, S. & Hayashi, H Sodefrin: a female-attracting peptide pheromore in newt cloacal glands. Science, 267, Kuramoto, M The amphibians of Iriomote of the Ryukyu Islands: ecological and zoogeographical notes. Bulletin of Fukuoka University of Education, 22 (Part III), Lannoo, J. M., Townsend, D. S. & Wassersug, R. J Larval life in the leaves: arboreal tadpole types, with special attention to the morphology, ecology, and behaviour of the oophagous Osteopilus brunneus (Hylidae) larva. Fieldiana Zoology, 38, Lefcort, H Adaptive, chemically mediated fright response in tadpoles of the southern leopard frog, Rana utricularia. Copeia, 1996, Liang, M.-F., Huang, C.-H. & Kam, Y.-C Effects of intermittent feeding on the growth of oophagous (Chirixalus eiffingeri)and herbivorous (Chirixalus idiootocus) tadpoles from Taiwan. Journal of Zoology, 256, Manteifel, Y Chemically-mediated avoidance of predator by Rana temporaria tadpoles. Journal of Herpetology, 29, SAS Institute SAS/STAT User s Guide. Cary, North Carolina: SAS Institute. Summers, K Paternal care and the cost of polygyny in the green dart-poison frog. Behavioral Ecology and Sociobiology, 27, Summey, M. R. & Mathis, A Alarm responses to chemical stimuli from damaged conspecifics by larval anurans: tests of three neotropical species. Herpetologica, 54, Thompson, R. L Larval habitat, ecology, and parental investment of Osteopilus brunneus (Hylidae). In: Contributions to West Indian Herpetology: a tribute to Albert Schwartz (Ed. by R. Powell & R. W. Henderson), pp Ithaca: Society for the Study of Amphibians and Reptiles. Ueda, H Reproduction of Chirixalus eiffingeri (Boettger). Scientific Reports of Laboratory of Amphibian Biology, Hiroshima University, 8, Waldman, B Olfactory basis of kin recognition in toad tadpoles. Journal of Comparative Physiology A, 156, Waldman, B Kin recognition in amphibians. In: Kin Recognition (Ed. by P. G. Hepper), pp Cambridge: Cambridge University Press. Wassersug, R. J., Frogner, K. J. & Inger, R. F Adaptations for life in tree holes by rhacophorid tadpoles from Thailand. Journal of Herpetology, 15, Weygoldt, P Complex brood care and reproductive behaviour in captive poison-arrow frogs, Dendrobates pumilio O. Schmidt. Behavioral Ecology and Sociobiology, 7, Wilson, E. O Sociobiology. Cambridge, Massachusetts: Harvard University Press.