Effects of hatching age on development and hatchling morphology in the red-eyed treefrog, Agalychnis callidryas

Biological Journal of the Linnean Society (1999), 68: 443–470. With 10 figures Article ID: bijl.1999.0325, available online at http://www.idealibrary.com on

Effects of hatching age on development and hatchling morphology in the red-eyed treefrog, Agalychnis callidryas KAREN M. WARKENTIN∗ Department of Zoology, University of Texas, Austin, TX 78712, U.S.A. Received 11 November 1998; accepted for publication 12 January 1999

The red-eyed treefrog, Agalychnis callidryas, lays eggs on leaves overhanging ponds. Tadpoles hatch and enter the water at different ages, and late-hatched tadpoles survive aquatic predators better than do early-hatched tadpoles. Here I assess developmental consequences of hatching age through: (1) a morphological study of embryos and tadpoles through the plastic hatching period; (2) a behavioural assay for an effect of hatching age on feeding; and (3) a field experiment testing the effect of hatching age on growth to metamorphosis. Substantial development of feeding, digestive, respiratory and locomotor structures occurs in embryos over the plastic hatching period. Hatchling morphology thus varies with age, with consequences for behaviour and predation risk. Hatched tadpoles develop faster than embryos, and early-hatched tadpoles feed before late-hatched tadpoles. After all tadpoles have hatched, the effect of hatching age on size decreases. I found no evidence for an effect of hatching age on size at metamorphosis and only weak evidence for an effect on larval period. Hatching age affects the sequence of developmental change: early-hatched tadpoles lose external gills while otherwise more developed embryos maintain them. Plasticity in external gill resorption may be adaptive given differences in the respiratory environments of embryos and tadpoles. Early-hatched tadpoles also diverge from embryos in shape, growing relatively smaller tails. The study of functional morphology and developmental plasticity will contribute to understanding hatching as an ontogenetic niche shift.  1999 The Linnean Society of London

ADDITIONAL KEY WORDS:—growth – respiration – predation – anuran – tadpole – differentiation – gills – phenotypic plasticity – ontogenetic niche. CONTENTS

Introduction . . . . . . Material and methods . . Morphology . . . . Onset of feeding . . . Field experiment: growth Results . . . . . . . Morphology . . . .

. . . . to . .

. . . . . . . . . . . . . . . . . . . . metamorphosis . . . . . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

444 446 446 448 448 449 449

∗ Present address: School of Biological Sciences, University of Kentucky, Lexington KY 40506-0225, U.S.A. E-mail: [email protected] 0024-4066/99/110443+28 $30.00/0

443

 1999 The Linnean Society of London

444

K. M. WARKENTIN

Onset of feeding . . . . . . . . . . . Field experiment: growth to metamorphosis . . Behavioural observations . . . . . . . . Discussion . . . . . . . . . . . . . . Hatchling morphology affects survival . . . . Hatchling morphology affects resource acquisition Short-term effects of hatching age on development Long-term effects of hatching age . . . . . Toward a developmental ecology of hatching . Acknowledgements . . . . . . . . . . . References . . . . . . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

458 459 461 461 462 463 463 465 466 467 468

INTRODUCTION

Changes in habitat, diet, predators or susceptibility to physical factors are common during development; such changes are called ontogenetic niche shifts (Werner & Gilliam, 1984). Life history theory predicts that shifts should occur when the ratio of risks to benefits in the alternative niche becomes more favourable than that in the current niche (Werner, 1986; Werner & Gilliam, 1984). Empirical studies of phenotypically plastic post-hatching niche shifts support the theory (e.g. amphibian metamorphosis: Crump, 1989; Newman, 1992; Skelly & Werner, 1990; sunfish habitat shifts: Werner et al., 1983a, b; Werner & Hall, 1988). Hatching clearly fits within the ontogenetic niche shift concept, and theory developed for post-hatching shifts has provided useful predictions about variation in hatching (Sih & Moore, 1993; Warkentin, 1995). Niche shift theory is largely size-based: it uses size-specific fitness correlates to predict the optimal size at transition points (Rowe & Ludwig, 1991; Werner, 1986; Werner & Gilliam, 1984). While an emphasis on size seems appropriate for posthatching niche shifts, for hatching, rapid developmental changes in morphology might be more important than size in determining effects of variation in the switch point on fitness. Considering only discrete niche shifts rather than gradual changes during development, there are two common types of post-hatching switches. The more dramatic is metamorphosis, wherein the ecological change is determined by morphological change: an anuran leaves the tadpole niche and enters the frog niche when it physically transforms. Animals enter post-metamorphic life with similar morphology, but at different sizes depending on pre-metamorphic growth. The other common shift is a change in habitat and/or diet during a period of simple growth, without change in form. Most fish experience such changes, as do many reptiles and invertebrates (reviewed in Werner & Gilliam, 1984). Again, animals enter the new niche with the same basic morphology but at different sizes. Thus in most ontogenetic niche shifts considered to date, size can vary substantially, while morphology varies little. Hatching is different. In the strict sense, measuring size by dry mass or energy content, embryos and pre-feeding hatchlings shrink. The increase in other measures of size (e.g. body length), achieved by absorbing water and transforming yolk into other tissue, is limited by the resources provided in the egg. Early ontogeny is, however, a period of great morphological change, and hatching can occur in very different morphological states. An extreme but illustrative case is poecilogonous

HATCHING AGE, MORPHOLOGY AND DEVELOPMENT

445

marine invertebrates such as the opisthobranch Haminoea callidegenita, in which swimming, pelagic veligers and crawling, benthic juveniles hatch from the same egg mass (Gibson & Chia, 1989). Morphology affects ecological interactions, including vulnerability to predators, ability to feed, and susceptibility to physical factors. Common size-dependent patterns may be reversed because of morphology. For example, under some conditions small hatchlings of the frog Bombina orientalis are less vulnerable to predators than are large hatchlings (Kaplan, 1992). The small hatchlings have relatively smaller yolk sacs, compared to tail size, and are better swimmers (Parichy & Kaplan, 1995). Size or age provides a convenient proxy for development in some cases (Sih & Moore, 1993; Warkentin, 1995); however, a more complete understanding of the ecology of hatching will require explicit consideration of morphological change and its effects. In the simplest cases, standardized stages of development might appropriately replace size. This may be inadequate, however, if environmental variation or the change of environments at hatching affects development. For instance, environmental effects on shape can alter vulnerability to predators (McCollum & Van Buskirk, 1996; Parichy & Kaplan, 1995; Smith & Van Buskirk, 1995). Environmental variation may have direct physical or chemical effects on development. The different egg and post-hatching environments may also have selected for different developmental trajectories in embryos and hatched animals. Knowledge of morphological change and its consequences in the egg and post-hatching environments through the period when hatching is possible is critical to understanding both the ecological effects of hatching age and the selective forces acting on hatching timing. Here I present the first detailed study of morphological change through the hatching period and effects of hatching age on development for a species with a phenotypically plastic hatching stage. I then discuss the relationship between morphological variation, behaviour, and the fitness consequences of hatching at different ages. My study organism, the red-eyed treefrog, Agalychnis callidryas, is an excellent species for examining hatching as a niche shift. The timing of hatching is plastic and mortality risks facing embryos and hatchlings are clearly distinct (Warkentin, 1995). Increased risk to embryos triggers early hatching, as predicted by niche shift theory, however, early-hatched tadpoles suffer higher predation by aquatic predators than do late-hatched tadpoles (Warkentin, 1995). As in Bombina orientalis, large tadpoles can be more vulnerable to predators than are smaller tadpoles, thus fitness correlates are not strictly size-dependent (Kaplan, 1992; Warkentin, 1998). A. callidryas do not feed immediately upon hatching, and the onset of resource acquisition after hatching depends on development. To better understand both why delayed hatching improves survival and if early hatching could offer any other fitness-related benefits, here I determine how morphology of A. callidryas tadpoles arriving in the water varies with hatching age, and assess effects of hatching age on feeding, growth and differentiation. I report the results of three experiments: (1) a morphological study of development through the plastic hatching period (age 5–8 d) in arboreal embryos versus aquatic tadpoles hatched at 5 d and raised in water, (2) a behavioural study of the effect of hatching age on the initiation of feeding, and (3) a field experiment on the effect of hatching age on growth through the larval period and metamorphosis.

446

K. M. WARKENTIN MATERIAL AND METHODS

Morphology I used a split-clutch design to determine what development occurs during the plastic hatching period (age 5–8 d; Warkentin, 1995), and assess any differences in development between the arboreal egg and aquatic post-hatching environments. The part of standard staging tables (e.g. Gosner, 1960) that applies to A. callidryas through the hatching period is based on opercular development and external gill loss. In A. callidryas, these changes are dissociated from other developmental events and associated with hatching (see Results), thus they do not provide a good general measure of development. Rather than rely on staging tables to characterize development I specifically examined a series of features of potential ecological and physiological importance.

Specimen collection I collected five clutches of young A. callidryas eggs from vegetation overhanging ponds near Sirena Biological Station, Corcovado National Park, Costa Rica, in 1993. I divided each clutch into two parts, hung the eggs over stream water in plastic containers, and misted them with water daily to prevent desiccation. At 5 d post-fertilization I mechanically induced hatching in one part of each clutch, and from each resulting set of sibling tadpoles I placed five in 1 liter of stream water (aquatic post-hatching treatment). I left the other part of each clutch to continue development in the egg (arboreal egg treatment). The containers of eggs and tadpoles (N=5 of each) were interspersed in close proximity in the shade in an open-air laboratory, and exposed to the same small natural fluctuations in temperature. When I induced hatching, I immediately preserved one animal from each clutch in 10% neutral buffered formalin. I preserved one newly hatched animal from the egg treatment (henceforth ‘hatchling’) and one previously hatched animal from the posthatching treatment (henceforth ‘tadpole’) from each clutch at daily intervals thereafter (N=5 specimens per age per treatment). All embryos had hatched from one clutch by 8 d so a hatchling from an additional, unrelated clutch was preserved instead. To avoid inducing hatching of animals in the egg treatment, I did not measure clutch temperatures. To assess if eggs are consistently warmer or cooler than tadpoles in water, at five times during the plastic hatching period I used a cloacal thermometer to measure temperatures of five clutches and five containers of water maintained under similar conditions in Gamboa, Panama, in 1998.

Morphometrics For each animal, I measured a series of external features including tail length and head-body length, tail muscle area and tail fin area in profile, maximum body width, head width at eyes, snout length (from snout tip to midway between the eyes in dorsal view) and length of external gill filaments. I viewed specimens under dark field illumination through a dissecting microscope and measured them with a computerized image processing system connected to the microscope via a video camera head.

HATCHING AGE, MORPHOLOGY AND DEVELOPMENT

447

Dissections I dissected a series of specimens to examine the viscera. In most cases I examined five specimens from each combination of age×environment (egg or post-hatching). The only exception was gut measurements of 8 d tadpoles and 8 d hatchlings, which I made on only four specimens, leaving the fifth intact for SEM examination of external features. I dissected and measured specimens under a dissecting microscope with an ocular micrometer. I first examined mouthparts unstained for evidence of keratinization, then stained them with crystal violet to see soft structures. I exposed viscera in ventral view and examined gut development, including growth, differentiation and gut filling. Tadpole intestines coil clockwise from the stomach anlage in a series of approximately circular loops. Part way along the intestine the direction of coiling abruptly reverses, to counter-clockwise, forming the switchback (Nodzenski, Wassersug & Inger, 1989). Coiling continues counter-clockwise to the anus. I measured gut diameter at a position 90° around the coil orally from the switchback, counted the number of intestinal coils (if any) oral and anal of the switchback, and measured the diameter of each coil. I estimated the length of coiled gut from the number of coils and their diameters, assuming circularity. I noted differentiation of the liver and pancreas. I removed the digestive system and liver from the body cavity to expose the lungs, and measured length and maximum width of each lung.

Scanning electron microscopy To further characterize development through the hatching period, I examined 5 d hatchlings, 8 d hatchlings and 8 d tadpoles hatched at 5 d using scanning electron microscopy. For each age and treatment, I prepared a single specimen as a whole body mount to view external structures, and dissected 2 specimens to expose internal oral structures. Prior to dissection, I injected crystal violet dye through the esophagus into the branchial and buccal cavities to visualize internal oral surfaces. I made incisions from the oesophagus along the dorsal edge of the branchial baskets and around the sides of the mouth to the corners of the beaks to separate dorsal and ventral structures. Specimens for SEM were postfixed in osmium tetroxide (Malick & Wilson, 1975), dehydrated in a graded ethanol series and critical point dried from liquid CO2. Dried specimens were sputter-coated with gold and examined using a Cambridge S150 Scanning Electron Microscope.

Statistical analysis I used a principal components analysis to summarize variation in size and shape, using square roots of area measurements to standardize dimensions. I tested the scores from the first two components with a multivariate analysis of variance for effects of age, environment (egg or post-hatching) and interaction. Significant interaction effects indicate divergence in the developmental trajectories of embryos and tadpoles. Five day old embryos are both the starting point for post-hatching development of early-hatched tadpoles and a stage in continued embryonic development (see Fig. 1). Data from these animals were, therefore, entered twice, once in each category, and degrees of freedom were corrected to represent the number of specimens. Statistical analyses were performed in Systat 5.2.1 (SYSTAT Inc., Evanston IL).

448

K. M. WARKENTIN

Onset of feeding To determine the age when feeding begins, and any effect of hatching age on feeding onset, I monitored tadpoles hatched at different ages for the first signs of feeding. This was not a split-clutch design. I collected young clutches as above and induced eggs to hatch at age 5, 6, 7 or 8 d, in the early evening. I raised siblings together in 1 liter of stream water in a plastic container, and fed them a suspension of powdered alfalfa ad libitum. I examined each container of tadpoles at least twice daily at roughly 12 h intervals for the presence of faeces, an unambiguous indicator of feeding. After the check for faeces I observed the tadpoles for a 5 min period and recorded behavioural indicators of feeding. These included biting at the water surface, container walls or floor and vigorous buccal pumping. Sibling groups contained 5 tadpoles each for those hatched at 5 to 7 d (N=10 sibships/age). For tadpoles hatched at 8 d, groups included all siblings remaining unhatched until that point (N=9 sibships, 3 of 3, 4 of 5 and 2 of 6 tadpoles). Field experiment: growth to metamorphosis To assess natural post-hatching growth rates and any long-term effects of hatching age, I split clutches and raised siblings hatched at different ages in cages in their natal pond in Corcovado National Park, Costa Rica. I collected five clutches of A. callidryas eggs laid on the night of July 19, 1994, from a pond near Sirena Biological Station, divided each into 4 parts, and maintained them as above. I induced hatching in one portion of each clutch at each of ages 5, 6, 7 and 8 d, between 14:00 and 15:00 h. From each clutch at each hatching age I haphazardly selected five hatchlings; the only exceptions were 2 clutches in which only 2 eggs remained unhatched at 8 d. I measured total length to the nearest 0.25 mm for each hatchling by laying it in a wet petri dish and using a plastic ruler held under the dish. I then placed each group of siblings into a cage in the pond within 1.5 h of hatching. Cages were 22.9 cm square by 45.7 cm high, constructed of fibreglass window screen supported by a frame of galvanized wire, and covered with a screen lid attached with velcro. I placed cages vertically in the pond so that the tadpoles had access to the entire water column, including the muddy substrate that sifted in through the cage bottom. The four cages for each clutch were grouped together, thus block and clutch effects are confounded; the experiment does not evaluate possible interclutch differences. Blocks of cages were spaced at roughly 1.5 m intervals along the edge of the pond, at equal depths. As the water level varied I moved cages toward and away from shore as necessary to prevent drowning or drying, maintaining water level between 2/3 and 1/3 cage depth whenever possible. Periodically I removed tadpoles from the cages with a dipnet and measured them as above. I initially measured them daily, then less frequently later in the larval period. When metamorphosis began, I checked cages daily for metamorphs. Animals with at least one exposed forelimb (developmental stage 42; Gosner, 1960) were removed from the cage, measured for snout-vent length (SVL), and released. I repeated this experiment with an additional five clutches laid October 14, 1994, but the pond dried completely before the tadpoles reached metamorphosis. I collected growth data for these tadpoles through age 20 d. There was some mortality over the course of the experiment. All tadpoles in two

HATCHING AGE, MORPHOLOGY AND DEVELOPMENT

449

cages died when a tapir knocked over their cages on 2 August, preventing access to air. There were four unexplained deaths. In addition, 13 tadpoles had tail injuries (1 in each of 7 cages, 2 in 3 cages), probably due to small shrimp (Macrobrachium americanum) inserting their claws through the cage mesh. I analysed growth from age 8 to 20 d in all 10 clutches with a repeated-measures analysis of variance. I used ln-transformed cage mean of total length as the response variable, excluding tadpoles with discernibly clipped tails from the calculation. To better understand the changing effect of hatching age, I then examined the size distribution at particular ages with main effects tests (Winer, 1971). I analysed individual metamorph data from the first five clutches with a means model ANOVA, due to missing cells, and tested main and interaction effects with user-defined contrasts (Milliken & Johnson, 1984). I tested the random factor, clutch or block, over model mean square error and the fixed factor, hatching age, over interaction MS as appropriate for a mixed model design (Sokal & Rohlf, 1995).

RESULTS

Morphology Size and shape Principal components analysis of the correlation matrix of external measurements shows that size differences account for most of the variation. The first principal component has high positive component loadings for all variables (Table 1), so may be understood as a measure of overall size; PC1 explains about 90% of the variation. The second principal component explains most of the remaining variation and may be understood as an axis of shape. Body width, tail length, and tail muscle area load most heavily on PC2 (Table 1). Thus animals with high PC2 scores have relatively narrow bodies, long tails, and large tail muscles, while animals with low PC2 scores are relatively wide bodied with smaller tail muscle area and shorter tails. Tail fin area loads only weakly on PC2, indicating differences in shape depend more on the relative amount of tail musculature than of tail fin area (Table 1). Multivariate analysis of variance of PC1 and PC2 reveals significant effects of age, environment, and age×environment interaction (Wilk’s Lambda=0.105, 0.287 and 0.193 respectively, all P<0.001). Univariate F tests indicate that both variables contribute to these effects. Not surprisingly, older animals were larger (Fig. 1A, age effect on PC1: F1,31=259.636, P<0.001). Growth rates were also substantially higher in the water than in the egg (Fig. 1A, interaction effect: F1,31=77.122, P<0.001). From age 5 to 8 d embryos grew from 10.5±0.21 to 11.9±0.29 mm total length (mean±SE throughout, N=5). Over the same time period hatched tadpoles grew to 14.7±0.10 mm (N=5). Growth over the plastic hatching period was allometric, as indicated by PC2. Animals changed in shape as they grew, and over time the shape of embryos and hatched tadpoles diverged (Fig. 1B, age effect: F1,31=58.007, P<0.001; interaction F1,31=14.849, P=0.001). Older embryos had higher PC2 scores, indicating relatively longer, more muscular tails and narrower bodies than younger embryos. In hatched tadpoles PC2 initially increased slightly with age, then declined, so 8 d hatchlings had higher scores than tadpoles (Fig. 1B).

450

K. M. WARKENTIN

Figure 1. Development of late-hatching arboreal embryos and early-hatched aquatic tadpoles of Agalychnis callidryas through the plastic hatching period. (A) Size, measured as the first principal component from a series of external measurements. Tadpoles grow more rapidly than embryos. (B) Shape, the second component; PC2 increases with greater tail length and tail muscle area, and decreases with greater body width. Embryos and tadpoles diverge in shape. (C) Lung length, (D) maximum length of external gills, and (E) length of the coiled gut. Lung growth, external gill regression, and the differentiation and growth of intestinal coils is faster in tadpoles than embryos. Data are means and SE of N=5 animals, except for gut length of 8 d hatchlings and tadpoles N=4. Where error bars are not visible the SE are smaller than the diameter of the data points.

T 1. Factor loadings from a principal components analysis of morphological measurements of Agalychnis callidryas tadpoles. PC1 measures size. PC2 is a measure of shape; tadpoles with high values have relatively narrow bodies and long muscular tails Measurement

PC1

PC2

Tail length Tail muscle area (square root) Tail fin area (square root) Head-body length Head width Body width Snout length

0.956 0.948 0.949 0.985 0.989 0.840 0.978

0.227 0.278 −0.090 −0.092 0.018 −0.531 0.127

90.352

6.332

% of variance explained

Lungs The lungs of young A. callidryas are basically cylindrical, with a constricted tip. They were always uninflated in new hatchlings and inflated in tadpoles. There was no significant difference between left and right lung lengths (paired t-test: t33=0.345, P=0.732) and lung size was not measurably asymmetric in many animals, so average

HATCHING AGE, MORPHOLOGY AND DEVELOPMENT

451

lung size for each animal was used for analysis. Lungs grew in length over the plastic hatching period (Fig. 1C, Page test of ordered alternatives, Siegel & Castellan, 1988: embryos P=0.01, tadpoles P<0.001). Most of this increase came in the first day; in embryos lungs lengthened by 45% between age 5 and 6 d. Lungs grew significantly faster in tadpoles than in embryos; there was no overlap in lung lengths from age 6 d onward (Fig. 1C, Wilcoxon signed ranks test, Z=3.408, P=0.001, 6–8 d old embryos vs. tadpoles). There was no evidence for change in lung width over this period in either environment (analysis of variance, all P>0.3). Lung width measurements were less accurate than length measurements due to longitudinal pleats in the lungs. No transverse pleats were observed. Digestive tract Considerable growth and differentiation of the gut occurred from age 5–8 days. Gut coils formed and elongated (Fig. 1E, Page test on length of coiled gut: embryos P<0.01, tadpoles P<0.001). This process occurred significantly faster in tadpoles than in embryos (Fig. 1E, Wilcoxon signed ranks test of coiled gut length, 6–8 day old embryos vs. tadpoles, Z=2.805, P=0.005). By 7 d there was no overlap in length of coiled gut. Five d embryos had a simple yolk sac filling most of their abdominal cavity. The yolk was cohesive and completely filled the sac. There was a short, straight empty tube lying dorsally, connecting the yolk sac with the anus. The stomach anlage and oesophagus were not distinct; rather, a short, undifferentiated tube connected the yolk sac with the bucco-pharyngeal cavity. This tube was densely packed with yolk except for a narrow clear central space. The liver and pancreas were distinct, but the tissues appeared similar. The gall bladder and bile duct were not evident under gross dissection. In 6 d hatchlings, the stomach anlage was still not distinct from the oesophagus, but the foregut tube was longer, and only loosely filled with yolk. In 4/5 specimens the yolk sac had not begun to differentiate into gut coils. In one specimen there was a short fissure in the yolk, extending posteriorly and sinistrally from the anterior dextral edge of the yolk sac, adjacent to the exposed portion of the liver. A similar fissure was evident in three of five 6 d tadpoles hatched a day earlier; two already had 2.5 gut coils, 0.9 and 1 mm in diameter. In most specimens there was some differentiation between stomach anlage and oesophagus; the stomach anlage was slightly thicker walled, with a larger lumen. In two specimens this part of the gut was empty and in the other three it contained some yolk granules. The pancreatic tissue was slightly less dense and more granular in appearance than the liver. The gall bladder was evident and the bile duct indicated in the mesentery, connecting with the gut at the oral end of the yolk sac or yolk-filled coils. Seven d hatchlings, like 6 d tadpoles, had little to no yolk in their foregut, and a distinguishable stomach anlage and oesophagus. One specimen had only an initial fissure in the yolk sac, but the rest had well defined gut coils (2.7±0.22 coils, 0.8±0.08 mm diameter, mean±SE, N=4). The liver and pancreas were grossly distinguishable in these hatchlings. They all had a visible gall bladder, some with green bile evident, and a robust bile duct. Some showed a green bile spot in the yolk. There was no space free of yolk in the gut coils; yolk was solidly packed and cohesive except in one specimen where it was looser at the bile spot. In 7 d tadpoles, after 2 d in the water, the duodenum was in some cases distinguishable from the

452

K. M. WARKENTIN

stomach anlage, on the basis of wall thickness and lumen diameter. There was no yolk in the foregut except a few granules in the duodenum of one animal, and one animal had two food particles in its oesophagus. There were 4.6±0.3 well defined gut coils, 0.5±0.03 mm in diameter (N=5). They were packed with yolk against the walls from the bile duct junction anally to the switchback; the anal coils contained less yolk, becoming emptier farther along the gut, with the straight hindgut and in some specimens the last coil essentially empty of yolk. The yolk-packed oral coils had a well defined central space that was either free of yolk or filled with a liquid yolk suspension. No green bile was evident, though the gall bladder was large and the bile duct robust. Eight d hatchlings had a well defined, empty, oesophagus and stomach anlage and a weakly defined, empty duodenum. There were 3.3±0.29 gut coils, 0.5±0.08 mm in diameter (N=4). The gut coils were completely packed with yolk from the bile duct junction to or slightly beyond the switchback. A small green spot of softened yolk was evident at the bile duct junction but there was no space free of yolk. The coils anal to the switchback were empty or mostly empty of yolk, and the straight hindgut was empty. Eight d tadpoles had more gut coils (5.9±0.26 coils, 0.4±0.01 mm diameter, N=4). The oesophagus, stomach anlage and duodenum were well defined, free of yolk, and contained food particles. The gut coils oral to the switchback had their walls packed with yolk, while anally from the switchback the gut was empty of yolk. Food particles were present through the entire gut (two tadpoles) or through the stomach and into the oral coils, inside the yolk, but not past the switchback (three tadpoles). External mouthparts Terms used to describe oral disc development follow Altig (1970) and Altig and Thibaudeau (1988). At 5 d, embryos had begun to develop their external mouthparts (Fig. 2A). All five tooth row ridges were evident, although the second anterior ridge (AR-2) was only represented by short segments in the corners of the mouth, and the first posterior ridge (PR-1) had a relatively broad medial gap. There was, however, no sign of labial teeth (denticles) on the ridges or of keratinization of the jaw sheath (beak). A single row of papillae was present in each corner of the oral disc. Ciliated epidermal cells were present on the surface of tooth row ridges and jaw. Cement glands were prominent, with well-developed secretory cells. By age 6 d, in hatchlings the oral papillae extended farther around the oral disc, although there was still a large posterior gap. The PR-1 gap had narrowed somewhat and the first signs of jaw sheath keratinization were evident in wet specimens as a thin brown edge along the centre of the beak. In some 6 d tadpoles the PR-1 gap had closed; in others it was near to closing. Jaw sheath keratinization was more advanced than in embryos of the same age, extending the full width of the beak in a narrow to moderate line. In some individuals, the cement glands had begun to regress down the center of the gland. Seven d embryos had extensive oral papillae; the anterior gap width was similar to that of mature tadpoles and the posterior gap was narrow (1–2 papillae wide). The PR-1 gap had closed or almost closed, as in 6 d tadpoles. Cement glands were still well defined, but lower than in 6 d embryos. Jaw sheaths were narrowly to moderately keratinized. Labial teeth had begun to keratinize in the first posterior row (P-1) and, in some embryos, medially in P-2, although they had not yet emerged

HATCHING AGE, MORPHOLOGY AND DEVELOPMENT

453

Figure 2. Oral disc development of Agalychnis callidryas through the plastic hatching period. The animal newly hatched at 8 d (B) is developmentally intermediate between the 5 d hatchling (A) and the 8 d tadpole hatched at 5 d (C). Development of marginal papillae (mp), formation and keratinization of jaw sheath ( js) and denticles (d), and atrophy of the cement glands (cg) lateral to the mouth is more rapid in the water than in the egg. Note also the abundance of ciliated cells in hatchlings (A & B) and their absence in the tadpole (C). Anterior and posterior tooth row ridges (ar-1, ar-2, pr-1, 2 and 3) have formed in 5 d embryos, but lack denticles. External nares (en) are included for orientation. Scale bars=200 lm.

through the skin. Seven d tadpoles had a mature tooth row ridge configuration, with no PR-1 gap. There were two rows of papillae in the corners of the oral disc, and the posterior gap was very narrow, less than one papillae wide in four of five specimens. Cement glands had lost height overall and regressed down the centre of the gland. Jaw sheaths were moderately to well keratinized. Denticle keratinization varied among specimens, from slight keratinization in P-1 only to keratinization of P-1 and P-2 denticles, weaker keratinization of A-1 and A-2, and slight keratinization of a few P-3 denticles. At 8 d, embryos had a mature tooth row ridge configuration and almost complete papillae, with two rows in the corners and a narrow to minimal posterior gap (Fig. 2B). Cement glands were low but still distinct. Jaw sheaths were well keratinized but did not yet show serrations. Denticle keratinization varied among specimens; in some individuals it was evident in all rows, although much weaker in A-2 and especially P-3. Others showed less keratinization. Denticles had begun to emerge through the skin in P-1 and a few in P-2 (Fig. 2B). Cilia were still present on tooth row ridges. Eight d tadpoles had complete tooth row ridges and essentially complete

454

K. M. WARKENTIN

oral papillae (Fig. 2C). They had heavily keratinized jaw sheaths with serrated edges. All denticle rows were keratinized, and denticle emergence was well advanced except in P-3. There were no cilia on the mouthparts. Cement glands were relatively atrophied; they were low with indistinct edges and had regressed down the centre of the gland. Internal oral and branchial anatomy Terminology follows Wassersug (1976). The oral anatomy of 5 d hatchlings was the most simple, and that of early-hatched 8 d tadpoles the most complex (Fig. 3). In dorsal view, the buccal cavity of 5 d hatchlings showed only two small, round lateral ridge papillae, considered by Wassersug (1980) to be part of the postnarial papillae series, and a few slight pustulations. There was a small, rounded median ridge. In addition to larger lateral ridge papillae, 8 d hatchlings had a pair of postnarial papillae, four buccal roof arena papillae, additional pustulations, and a more pointed median ridge. Papillae were more developed in 8 d tadpoles, and the edge of the median ridge was more complex. Hatchlings had clumps of cilia in the pre- and postnarial arenas, but 8 d tadpoles had none. In ventral view, 5 d hatchlings had a pair of small, rounded lingual papillae, a pair of low infralabial papillae, and a few pustulations indicating where buccal floor arena (BFA) papillae were beginning to form. Eight d hatchlings had larger infralabial and lingual papillae, six distinct BFA papillae, and additional pustulations. Eight d tadpoles had slightly longer papillae, more numerous pustulations, and larger, better defined buccal pockets. The margin of the ventral velum was also more elaborate. The differences in branchial basket development were more striking than differences in the buccal cavity (Figs 3 & 4). Five d hatchlings had straight gill arches (Fig. 3A). Gill filters were beginning to form, indicated by simple, short ridges separated by shallow grooves (Figs 3A & 4A). Neither gill arches nor forming filter rows were curved. Thus the forming filter chambers were simply narrow grooves between adjacent filter plate anlagen (Figs 3A & 5A). In 8 d hatchlings gill arches were curved. There were more developing filter rows per gill arch, and filter rows were over twice as long, rougher in texture, and curved (Figs 3C & 4B). Filter chambers were slightly larger, although still shallow. Compared with 8 d hatchlings, in 8 d tadpoles the branchial baskets were almost 3 times larger in projected area, with much deeper filter chambers and more filter rows (Fig. 3E). Secondary filter folds were present on gill filters except at dorsal edges of filter plates. As well, open apertures indicative of apical secretory cells (Viertel, 1991) were evident on the gill filters (Fig. 4C). External gills All hatchlings had long, branched filamentous external gills (Fig. 5A); hatching occurred at Gosner (1960) stage 23. In 5 and 6 d hatchlings the gills extended back from the operculum at least as far as the body-tail junction; in all but one of the 7 and 8 d hatchlings the gills were slightly shorter. In the 7 d hatchlings, the opercular opening had begun to narrow; the right external gill thus emerged ventrally rather than laterally as in younger hatchlings. In some specimens the ventrally located right gill filaments were interspersed with left gill filaments, thus I report maximum gill length without separating right and left gills. In one 8 d hatchling the right gill filaments were mostly covered by the operculum, with the left filaments still protruding

HATCHING AGE, MORPHOLOGY AND DEVELOPMENT

455

Figure 3. Bucco-pharyngeal cavity development of Agalychnis callidryas through the plastic hatching period. Ventral and dorsal views of newly hatched embryos aged 5 d (A, B) and 8 d (C, D), and an 8 d old tadpole hatched at 5 d (E, F). The cracks in (A) and (D) and fragmentation in (F) are drying artifacts. Branchial baskets (bb) grow and differentiate during the hatching period, as do the papillae in the buccal cavity. Note the differentiation of buccal floor arena papillae (bfap) and the papillae posterior to the internal nares (in). Scale bars=400 lm.

456

K. M. WARKENTIN

Figure 4. Gill filter development of Agalychnis callidryas through the plastic hatching period. Left third gill arch of newly hatched embryos aged 5 d (A) and 8 d (B), and an 8 d old tadpole hatched at 5 d (C). Note the increase in length and the curvature of filter row anlagen from 5–8 d in embryos (A, B). Surface complexity is greatly increased in the 8 d tadpole (C); secondary filter folds and the open apertures of secretory cells are evident. Scale bars=40 lm.

HATCHING AGE, MORPHOLOGY AND DEVELOPMENT

457

Figure 5. External gill filaments of Agalychnis callidryas hatchlings. (A) Lateral view of the anteroventral portion of a 5 d hatchling, showing lungs, developing branchial baskets, and external gill filaments. The box indicates a portion of a single gill filament shown magnified in (B). (C) Portion of a single gill filament of an 8 d hatchling, for comparison. Scale bars: 400 lm in (A) and 10 lm in (B) and (C).

fully; three had both gills exposed. The gills of one specimen were missing, apparently broken in preparation. The maximum length of external gills declined with age (Fig. 1D, ANOVA: F1,17=14.192, P=0.002). The gill filaments were covered with dense rows of cilia and appeared similar in 5 and 8 d hatchlings under SEM (Fig. 5B and C). Tadpoles did not have external gills. One 6 d tadpole had a few gill filaments

458

K. M. WARKENTIN

Figure 6. Neuromasts on the tail of a 5 d Agalychnis callidryas hatchling. Scale bar=10 lm.

protruding less than 0.25 mm out of the spiracle (stage 24, Gosner, 1960). The rest had no visible external gills (stage 25), and their opercular openings were flat and less than 0.5 mm wide. Opercular openings of 7 and 8 d tadpoles were narrower still, and there was no sign of gill filaments under the operculum in dissection. Lateral line system Five and 8 d hatchlings and 8 d tadpoles had externally well-developed neuromasts (Fig. 6). No attempt was made to quantify lateral line development, but there were no obvious differences in the morphology or distribution of neuromasts among the specimens examined. Development temperature Average clutch and water temperatures were never more than 1°C different. In three morning observations, eggs were slightly warmer than the water (Mann– Whitney U, P=0.008, 0.007 and 0.02). In two evening observations, temperature did not differ significantly between eggs and water (Mann–Whitney U, P=0.82 and 0.18). I have, on other occasions, recorded slightly higher water than clutch temperatures shortly after a drop in air temperature during rain. Onset of feeding Tadpoles did not begin feeding immediately after hatching. Animals that hatched earlier took significantly longer after hatching to begin feeding, as indicated by

HATCHING AGE, MORPHOLOGY AND DEVELOPMENT

459

Figure 7. Timing of initiation of feeding in Agalychnis callidryas tadpoles hatched at different ages. Data are proportion of sibships first producing faeces at each age. Age is measured from the evening of oviposition. N=10 sibships per age except N=9 sibships at age 8 d.

faeces production (Fig. 7, Jonckheere test of ordered alternatives, Siegel & Castellan, 1988: J∗=4.51, P<0.001). Nonetheless, tadpoles hatched early did begin feeding significantly earlier than tadpoles hatched later (Fig. 7, Jonckheere test: J∗=5.44, P<0.001). Not surprisingly, feeding behaviour was evident before faeces were produced. The first biting was usually seen one or two observation periods before the first faeces (0.8±0.06 d earlier, mean and SE, N=39).

Field experiment: growth to metamorphosis At 1 d post-hatching tadpoles were in stage 25 (Gosner, 1960); they lacked grossly visible external gills. Although I cannot rule out the presence of microscopic remnants of external gills, the external gills in newly hatched tadpoles were conspicuous to the naked eye. As in the laboratory, early-hatched tadpoles raised in cages in their natal pond grew faster than their embryonic siblings, giving them an initial size advantage over late-hatched animals (Fig. 8). There was, however, a significant age×hatching age interaction; once all animals were in the water the size difference among tadpoles hatched at different ages decreased over time (Fig. 8, repeated measures ANOVA, interaction effect: F18,204=9.238, Greenhouse–Geisser corrected P<0.001). Thus at 8 d hatching age strongly affected size (main effects test: F3,36=60.395, P<0.001) but by age 14 d, when the latest hatched tadpoles had been in the water for 6 d, there was no longer a significant effect of hatching age on size (main effects test: F3,34= 2.028, P=0.129). Animals hatched at 6 and 7 d were similar in size to earlierhatched tadpoles after only a day in the water (Fig. 8). No main effects test on tadpoles older than 14 d showed a significant effect of hatching age on size. Froglets ranged from 15.5 to 18.5 mm SVL at forelimb emergence (17.3±0.07 mm, mean and SE, N=80). Neither hatching age nor its interaction with clutch/block significantly affected SVL at forelimb emergence (F3,10=0.554, P=0.657 and F10,62=1.626, P=0.120, respectively). Clutch or block affected SVL at metamorphosis (F4,62=2.760, P=0.035). Most froglets metamorphosed from 59 to 83 d after oviposition (69±0.69 d, N= 78, Fig. 9). Two ‘runts’ metamorphosed at 90 and 92 d, 20 and 18 days after the mean

460

K. M. WARKENTIN

Figure 8. Growth of Agalychnis callidryas tadpoles hatched at different ages and raised in a natural pond. Data are means of cage means and SE. N=10 sibships except as follows. At age 14 d, for tadpoles hatched at 5 and 8 d, N=9 sibships. At age 60 d, for tadpoles hatched at 6 and 7 d, N=5 sibships, and for tadpoles hatched at 5 and 8 d, N=4 sibships.

Figure 9. Age at metamorphosis of Agalychnis callidryas tadpoles hatched at different ages and raised in a natural pond. Data are means and SE for siblings sharing a cage. Data points for siblings hatched at different ages are connected; unconnected data points are outliers not included in the cage mean and SE. N=5 tadpoles/cage except as follows. For sibship 1 hatched at 5 d, N=4. For sibship 2 hatched at 8 d, N=2. For sibship 3 hatched at 7 d, N=3, and hatched at 8 d, N=2. For sibship 5 hatched at 5 d, N=4.

date of metamorphosis of other tadpoles in their cages (Fig. 9). These statistical outliers (Studentized residuals >3) made the data both heteroscedastic and non-normal. Natural-log transformation of the full data set, including outliers, controlled the heteroscedasticity (Bartlett’s test: P>0.1) but failed to correct the non-normality (Kolmogrov–Smirnov test: P=0.04). Removing the outliers corrected both non-normality and heteroscedasticity (Kolmogrov–Smirnov test: P=0.484; Bartlett’s test:

HATCHING AGE, MORPHOLOGY AND DEVELOPMENT

461

P>0.9). Analyses of variance on both transformed and reduced data sets failed to detect a significant interaction effect on age at metamorphosis (ln-transformed: F10,62=0.613, P=0.797; outliers removed: F10,60=0.961, P=0.484). Analyses of both data sets detected a highly significant effect of clutch or block on age at metamorphosis (lntransformed: F10,62=10.354; outliers removed: F4,60=11.786, both P<0.001). The effect of hatching age on age at metamorphosis was weakly significant (ln-transformed: F3, 10=4.646, P=0.028) or marginally non-significant (outliers removed: F3,10=3.133, P=0.074). On average, tadpoles hatched at 7 d transformed earliest (67.2±1.3 d, mean of cage means, N=5 cages) and those hatched at 5 and 8 d transformed later (Fig. 9, 72.2±2.6 and 72.5±3.3 d respectively, both N=4 cages, outliers included).

Behavioural observations Embryos 3 d and older were curled within the egg. From 4 d, one side of the body and tail lay against the vitelline membrane, the other surrounded a central space filled with perivitelline fluid. One external gill often lay between the body and the vitelline membrane, close to the outside air. Embryos usually abducted their other gill so that it extended away from the body and closer to the egg surface. Tadpoles fell to the bottom of the water upon hatching, but soon surfaced and filled their lungs with air. Five d hatchlings, even with full lungs, consistently sank unless they actively swam or adhered to a surface. Older hatchlings with air in their lungs sometimes maintained a midwater position or slowly rose without any tail movement, indicating neutral or positive buoyancy. Tadpoles often adhered to the surface film of the water by surface tension, hanging passively from their oral discs. Animals 6 d and older consistently succeeded in adhering to the surface film on their first attempt. However, I often observed 5 d hatchlings swim to the surface, attempt to adhere, but fall down immediately. They also sometimes adhered briefly and then fell. In contrast they consistently adhered to surfaces within the water using their cement glands.

DISCUSSION

Young Agalychnis callidryas hatchlings have less chance of surviving in the water than do older hatchlings (Warkentin, 1995; 1999). They are also unable to feed, so receive no immediate benefits from the resources that become accessible after leaving the egg. Both risks and benefits in the post-hatching environment are affected by the morphological state in which tadpoles enter that environment, which varies with hatching age (summarized in Fig. 10). Although later hatchlings are larger than earlier hatchlings, the increase in size cannot alone explain the fitness differences. Hatching also affects growth and differentiation; embryos are morphologically distinct from equal-aged tadpoles that hatched earlier. The different development of embryos versus tadpoles may reflect both direct effects of their environments and adaptive, plastic responses. I will first address the immediate effects of hatchling morphology on survival and feeding in the water. I then examine the short- and long-term consequences of hatching age for growth and differentiation, and their

462

K. M. WARKENTIN

Figure 10. Morphological change in Agalychnis callidryas embryos and tadpoles through the plastic hatching period. Five d is the earliest that tadpoles hatch, and few embryos delay hatching past 8 d. A comparison between 5 d and 8 d hatchlings and 8 d tadpoles hatched at 5 d shows the relative differences in embryonic and post-hatching development through the plastic hatching period. Drawings are simplified representations of selected anatomical features. Body size, length of gill filaments and lungs, and number of gut coils represent average measured differences among animals. Differences in tail length and muscle area are exaggerated by a factor of 2 for visibility. Branchial baskets are drawn from individual scanning electron micrographs. Oral disks are drawn from scanning electron micrographs and wet dissection notes; jaw sheath serrations and denticle size are exaggerated for visibility.

relationship to fitness. Finally, I discuss how greater attention to morphology can inform our understanding of hatching as a life history event.

Hatchling morphology affects survival Older hatchlings are better than younger ones at coping with predaceous shrimp (Warkentin, 1995). This difference in survival is due to morphological developmental differences, rather than experience, since all hatchlings are equally inexperienced in the water. Age-related behavioural changes clearly contribute to improved survival. In the presence of shrimp, older hatchlings avoid the bottom more than do younger hatchlings (Warkentin, 1998, 1999). Reduced presence on the bottom enhances survival, since shrimp usually forage on the bottom and are more successfully at catching tadpoles there (Warkentin, 1998, 1999). Several morphological changes appear to contribute to this critical change in hatchling behaviour. First, lung growth affects buoyancy and the energetic cost of staying off the bottom. With their relatively large lungs inflated, hatchlings 6 d and older achieve neutral buoyancy but 5 d hatchlings, with smaller lungs, consistently sink (Figs 1 & 10). Secondly, tadpoles use their external mouthparts to adhere to the water surface. From 6 d, hatchlings are competent to adhere. Five d hatchlings, with less developed external mouthparts, sometimes fail to adhere and fall down (Figs 2 & 10). Lower buoyancy cannot completely account for this, since these animals readily adhere to surfaces within the water using their cement glands (Fig. 2). Finally, the relatively longer, more muscular tails of older hatchlings (Figs 1 & 10) could improve swimming performance, as has been demonstrated in Bombina orientalis hatchlings (Parichy &

HATCHING AGE, MORPHOLOGY AND DEVELOPMENT

463

Kaplan, 1995). This could enhance the ability to maintain a mid-water position and so avoid benthic predators. Larger tails could also improve escape ability, as suggested for B. orientalis (Kaplan, 1992; Parichy & Kaplan, 1995) and Pseudacris triseriata (McCollum & Van Buskirk, 1996). Hatchling morphology affects resource acquisition No A. callidryas begin feeding immediately after they hatch, but the delay is up to three times longer for early-hatched versus late-hatched animals. The immature state of feeding and digestive structures in early hatchlings is the apparent proximate cause of the longer delay (Figs 1–4 & 10). Agalychnis callidryas tadpoles are microphagous suspension feeders. They primarily use gill filters and pharyngeal mucus to trap food particles, while buccal papillae contribute to particle sorting (Kenny, 1969; Viertel, 1985, 1989; Wassersug, 1972, 1980). While the feeding structures of 8 d early-hatched tadpoles clearly function, the oral disc and bucco-pharyngeal cavity of 5 d hatchlings are substantially less developed. At 5 d the branchial baskets show neither the structural complexity nor the mucus secretion capacity necessary to trap small particles (Figs 2 & 3). Older hatchlings are morphologically intermediate, thus developmentally closer to foraging competence than young hatchlings (Figs 2 & 3). Moreover, the digestive system in the youngest hatchlings has barely begun to form from the yolk sac. Thus even if they could trap food particles they could not digest them. In contrast, in the oldest hatchlings the digestive system is well-formed, although still completely filled with yolk for part of its length (Figs 1 & 10). Short-term effects of hatching age on development Increase in rate of development A striking effect of early hatching is the overall increase in the rates of growth and differentiation relative to embryos of the same age (Figs 1–4, 8 & 10). From age 5–8 d, tadpoles increase in length 2–3 times faster than embryos. Differentiation of mouthparts, branchial baskets, and gastrointestinal tract and growth of lungs is likewise faster (Figs 1–4 & 10). This effect is evident in tadpoles raised both in the laboratory and in cages in a natural pond. All my observations on A. callidryas hatched at different ages for a variety of experiments over 5 years of research are consistent with this result; the pattern is thus not an artifact of any particular set of rearing conditions. Why do tadpoles develop faster after hatching? Natural selection can act on development rate, but there is no obvious selective advantage to slower development of embryos than tadpoles when hatching may occur at any moment. Rather, I think the change in environment at hatching probably affects growth and differentiation rates directly. I will consider three alternative factors: food, temperature, and gas exchange. Feeding by hatched tadpoles of course increases their growth relative to non-feeding embryos. However, growth and differentiation rates increase immediately upon hatching, well before feeding begins, and thus cannot be explained by feeding (Figs 1 & 7). Since the acceleration in development at hatching occurs in pre-feeding animals, it reflects a change in the rate of yolk conversion into tadpole tissue, indicating a

464

K. M. WARKENTIN

higher metabolic rate of tadpoles than embryos. Both temperature and gas exchange affect metabolism and rate of embryonic development in amphibians (Bradford & Seymour, 1988; Duellman & Trueb, 1986; Seymour & Bradford, 1995). Warmer water than clutch temperatures could explain faster growth and differentiation after hatching. I do not have temperature data for the embryos and tadpoles in the morphometric study, but temperature data from other clutches raised under similar conditions do not support the hypothesis that tadpoles are warmer after hatching. The third possibility is that oxygen availability limits metabolism, and so development rate, of embryos relative to tadpoles. Tadpoles have greater gas exchange opportunities than embryos. Both embryos and tadpoles use skin and gills for gas exchange with an aqueous medium; however, oxygen reaching embryos must first diffuse through the egg capsule. Tadpoles also use lungs in respiration. Post-hatching increases in oxygen consumption have been reported in other amphibians (Bradford & Seymour, 1985; Burggren, Infantino & Townsend, 1990). Oxygen limitation can cause slowed or arrested embryonic development (Booth, 1995; Bradford & Seymour, 1988; Cohen & Strathmann, 1996; Seymour, Mahony & Knowles, 1995). This effect is most likely for egg clutches in still water, where boundary layers of oxygen-depleted medium form. However, even single eggs in air can be oxygen-limited (Seymour, Geiser & Bradford, 1991). This is particularly likely for large, warm eggs because of their higher oxygen demand (Seymour & Bradford, 1995). Agalychnis callidryas embryos are large and warm; air temperature at Sirena is usually 25–27°C in the wet season (pers. obsv.). As well, A. callidryas embryos are surrounded on several sides by other respiring embryos, reducing the egg surface area available for diffusion. While certainly adequate to support development, oxygen concentration within the eggs may not support maximal development rates. This hypothesis could be tested by manipulations of oxygen availability and measurements of metabolic, growth and differentiation rates in preand post-hatching animals. Divergence in timing and developmental trajectory Some effects of early hatching appear to be a simple speeding up of growth and differentiation of tadpoles, along an essentially similar path as embryonic development. Early hatching also changes the developmental trajectory: the timing of gill resorption is altered, and shape diverges. These effects of hatching age on development could be adaptive responses to differences between embryonic and post-hatching environments. The stage at which external gills are resorbed varies with hatching age, with early-hatched tadpoles losing gills by 6 d and late-hatching embryos retaining gills largely unchanged as long as they are in the egg (Figs 5 & 10). This can be as late as 10 d (Warkentin, 1995). Meanwhile, development of other structures continues (Fig. 10). Gill resorption could be a simple response to increased oxygen availability, but its timing in association with hatching and its relative independence from other developmental events probably enhances fitness. For embryos, external gills may be necessary for adequate gas exchange. Not only do they provide a respiratory surface but also, unlike internal gills, external gills can be positioned to minimize diffusion distance to the air. Indeed, the maintenance of embryonic gills may facilitate delayed hatching, and so avoidance of aquatic predators, since oxygen stress is a common trigger of hatching (DiMichele & Taylor, 1980; Petranka et al., 1982; Bradford & Seymour, 1988).

HATCHING AGE, MORPHOLOGY AND DEVELOPMENT

465

Hatched tadpoles can ventilate internal gills with water and breathe air. This additional respiratory surface appears to make external gills unnecessary. Moreover the long, delicate external gill filaments (Fig. 5) may both reduce swimming performance by creating drag and be susceptible to damage in a pond environment. Their resorption by hatched tadpoles may thus be beneficial. Embryos change in shape over time, growing relatively slimmer with longer and more muscular tails. Hatched tadpoles also change in shape but—unlike overall growth or the differentiation of feeding and digestive structures—the shape changes in tadpoles are not simply a faster version of shape changes in embryos (Figs 1 & 10). The divergence in shape may be an adaptive (sensu Travis, 1994) response to the differences between egg and water environments. There are two factors to which this pattern of shape changes might be a response. First, the relatively larger tails of embryos may function to increase surface area for gas exchange. Since hatched tadpoles have access to air, they may be less dependent than embryos on cutaneous respiration. Second, larger tails may serve an anti-predator function. Some tadpoles respond to increased predation risk by growing relatively larger tails, which are thought to improve escape ability (McCollum & Leimberger, 1997; McCollum & Van Buskirk, 1996; Smith & Van Buskirk, 1995). Studies of Hyla and Pseudacris have demonstrated changes in both tail fins and tail musculature. Tail musculature clearly affects swimming performance (Hoff, 1987). The early-hatched tadpoles on which morphometrics were measured were raised in a predator-free environment. They may have diverted resources from the development of defensive structures to other functions. Embryos probably lack information about predation risk in their future aquatic environment, but A. callidryas breeding ponds often are predator-rich which could select for a default defensive strategy. In support of the second hypothesis, I show elsewhere that early-hatched tadpoles raised with predators, like older embryos, develop relatively large tails, and earlyhatched tadpoles raised without predators become more—not less—vulnerable to predatory shrimp with increasing age (Warkentin, 1998). This also demonstrates the limitations of size alone in predicting fitness differences at this stage. Since earlyhatched tadpoles grow faster, a size-based view would predict that they should survive with predators increasingly better than equal-aged new hatchlings. Even if raised with predators, however, early-hatched tadpoles do not survive better than equal-aged hatchlings when exposed to shrimp (Warkentin, 1998).

Long-term effects of hatching age Accelerated post-hatching growth and the value of early hatching Both a faster growth rate and earlier feeding may be considered advantages of early hatching. Based on the simple criterion of minimizing the mortality risk to growth rate ratio (l/g; Werner, 1986; Werner & Gilliam, 1984), we would expect earlier hatching under these conditions than if hatching did not alter growth. However, the initial size advantage of early-hatched tadpoles is transient (Fig. 8). Thus any competitive advantage or lowered predation risk associated with greater size is also transient. Indeed, even in the short-term, the larger size of early-hatched tadpoles does not enhance survival compared with small new hatchlings (Warkentin,

466

K. M. WARKENTIN

1998). Faster post-hatching growth therefore does not mitigate the predation risk cost of early hatching. As would be expected from the diminishing effect of hatching age on size over time, I found no evidence for effects of hatching age on size at metamorphosis, 2 months later. The long-term effects of hatchling characteristics can depend on the quality of the larval environment (Parichy & Kaplan, 1992), so there may be conditions under which hatching age affects size at metamorphosis. Nonetheless, my data suggest that under natural conditions any long-term effects of hatching age on size are small compared with early effects, and the immediate post-hatching increase in growth rate may have little ecological or evolutionary importance. Timing of metamorphosis There may be a weak effect of hatching age on age at metamorphosis. On average, tadpoles hatched at intermediate ages metamorphose sooner and the earliest- and latest-hatched animals transform slightly later. Because pond drying kills some A. callidryas tadpoles, late metamorphosis may carry a high fitness cost. Some amphibian larvae, however, respond to pond drying by hastening metamorphosis (reviewed in Newman, 1992). It is unknown if, or to what extent, A. callidryas exhibits such adaptive plasticity in metamorphosis. Thus the biological importance of an average 5 d difference in age at metamorphosis in a stable pond is equivocal. If the variation in age at metamorphosis is biologically significant, the selection it imposes on hatching age appears to be stabilizing. My data provide no evidence that size or age at metamorphosis imposes directional selection for earlier hatching. The most important selective forces acting on hatching age in A. callidryas are probably the mortality risks concentrated in a short time period immediately surrounding hatching, especially predation of eggs and tadpoles. This may be true for other species as well, especially species with relatively long larval periods or delays before sexual maturity (e.g. Petranka et al., 1987). Concentration of the effects of hatching stage in embryonic and early larval life would make them particularly amenable to study, requiring only short-term experiments.

Toward a developmental ecology of hatching Following niche shift theory, the best time to hatch should depend on the ratio of costs to benefits in each environment (Werner, 1986; Werner & Gilliam, 1984). The standard measure of costs, mortality rate, is clearly appropriate for embryos and hatchlings. Using growth rate to measure benefits, however, is problematic for pre-feeding organisms. Since development clearly affects the abilities to both survive and accrue resources in any particular environment, development rate may be preferable to growth rate for pre-feeding organisms, and stage-specific rates may be more useful than size-specific rates. Mortality and development rates in the egg and post-hatching environments depend on environmental factors such as predator density, oxygen availability and temperature. They also depend on the traits that the embryo or hatchling has developed: its morphology, physiology and behaviour. Considering, for simplicity, an invariant developmental trajectory, embryos should respond to egg

HATCHING AGE, MORPHOLOGY AND DEVELOPMENT

467

and post-hatching conditions based on their stage of development. For instance, since post-hatching survival improves with development (Sih & Moore, 1993; Warkentin, 1995; Moore, Newton & Sih, 1996), the perceived risk to embryos required to induce hatching should decrease with development. This is the case for A. callidryas (Warkentin, unpublished data). Factors that affect either development rate or survival should affect hatching stage. For instance, a reduction in embryonic development rate should favour earlier hatching. Indeed, submergence of terrestrial eggs reduces oxygen availability which both slows development and induces hatching (DiMichele & Taylor, 1980; Bradford & Seymour, 1988; Petranka, Just & Crawford, 1982). Development, however, is plastic. Both development rate (e.g. Wourms, 1972; Bradford & Seymour, 1985, 1988) and the trajectory of development (e.g. McCollum & Van Buskirk, 1996) can show adaptive plasticity in response to environmental variation. They are also directly affected by physical and chemical characteristics of the environment. Some environmental variables alter developmental trajectory in ways that affect stage-specific post-hatching predation risk (e.g. temperature in Bombina orientalis; Kaplan, 1992). Such changes in development should alter the optimal hatching stage under a given set of environmental conditions. In a proximate sense, development limits the range of variation in hatching to the period when embryos are capable of breaking out of the egg, but still capable of surviving within it. Developmental plasticity (e.g. delayed external gill loss) may extend the period when hatching is possible. If hatching stage varies, plasticity may also enhance the ability of embryos and larvae to survive and develop in their different environments. Developmental plasticity associated with variation in hatching time may provide clues to both the mechanistic basis and the evolution of hatching plasticity. Phenotypically plastic hatching is now recognized as a useful and interesting system in which to study the ecology and evolution of a life-history switch point (Sih & Moore, 1993; Warkentin, 1995; Moore, Newton & Sih, 1996; Blaustein, 1997). The evolutionary ecology of hatching should consider the ecological effects of the rapidly changing morphology of embryos and hatchlings, and any developmental plasticity they exhibit.

ACKNOWLEDGEMENTS

I thank M. Ryan, R. Wassersug, R. Dudley and D. Cannatella for advice and comments on the manuscript. The manuscript was also improved by comments from R. Kaplan, R. Buskirk, A. Sih and C. Gabor. I thank P. Valverde, G. Friesen and J. Poffenroth assistance in the field, R. Wassersug and S. Whitefield for assistance with dissections and SEM, and J. Young, G. Gage and B. Reiss for help in figure preparation. Laboratory facilities at Sirena were made possible by an NSF grant to L.E. Gilbert. I was supported during this work by fellowships from the University of Texas, and during manuscript revisions by a fellowship from the University of Kentucky. The research was funded by the UT Department of Zoology and Institute of Latin American Studies, a Gaige Award from the American Society of Ichthyologists and Herpetologists, the National Academy of Sciences through Sigma

468

K. M. WARKENTIN

Xi and an NSF dissertation improvement grant. The Department of Anatomy and Neurobiology at Dalhousie University provided SEM and image analysis facilities. This work was conducted under permits from the National Park Service and the Ministry of Natural Resources, Energy and Mines of Costa Rica, and approved by the University of Texas Animal Care Committee.

REFERENCES

Altig R. 1970. A key to the tadpoles of the continental United States and Canada. Herpetologica 26: 180–207. Altig R, Thibaudeau DG. 1988. Sequence of ontogenetic development and atrophy of the oral apparatus of six anuran tadpoles. Journal of Morphology 197: 63–69. Blaustein L. 1997. Non-consumptive effects of larval Salamandra on crustacean prey: can eggs detect predators? Oecologia 110: 212–217. Booth DT. 1995. Oxygen availability and embryonic development in sand snail (Polinices sordidus) egg masses. Journal of Experimental Biology 198: 241–247. Bradford DF, Seymour RS. 1985. Energy conservation during the delayed-hatching period in the frog Pseudophryne bibroni. Physiological Zoology 58: 491–496. Bradford DF, Seymour RS. 1988. Influence of environmental PO2 on embryonic oxygen consumption, rate of development, and hatching in the frog Pseudophryne bibroni. Physiological Zoology 61: 475–482. Burggren WW, Infantino RLJ, Townsend DS. 1990. Developmental changes in cardiac and metabolic physiology of the direct-developing tropical frog Eleutherodactylus coqui. Journal of Experimental Biology 152: 129–147. Cohen CS, Strathmann RR. 1996. Embryos at the edge of tolerance: effects of environment and structure of egg masses on supply of oxygen to embryos. Biological Bulletin 190: 8–15. Crump ML. 1989. Effect of habitat drying on developmental time and size at metamorphosis in Hyla pseudopuma. Copeia 1989: 794–797. DiMichele L, Taylor MH. 1980. The environmental control of hatching in Fundulus heteroclitus. Journal of Experimental Zoology 214: 181–187. Duellman WE, Trueb L. 1986. Biology of Amphibians. New York: McGraw-Hill. Gibson GD, Chia F-S. 1989. Developmental variability (pelagic and benthic) in Haminoea callidegenita (Opisthobranchia: Cephalaspidea) is influenced by egg mass jelly. Biological Bulletin 176: 103–110. Gosner KL. 1960. A simplified table for staging anuran embryos and larvae with notes on identification. Herpetologica 16: 183–190. Hoff K. 1987. Morphological determinants of fast-start performance in anuran tadpoles. Unpublished D. Phil. Thesis, Dalhousie University. Kaplan RH. 1992. Greater maternal investment can decrease offspring survival in the frog Bombina orientalis. Ecology 73: 280–288. Kenny JS. 1969. Pharyngeal mucous secreting epithelia of anuran larvae. Acta Zoologica 50: 143–153. Malick LE, Wilson RB. 1975. Evaluation of a modified technique for SEM examination of vertebrate specimens without evaporated metal layers. In: Johari, O, Corvin, I, eds. Scanning Electron Microscopy. Chicago: IIT Research Institute, 259–266. McCollum SA, Leimberger JD. 1997. Predator-induced morphological changes in an amphibian: predation by dragonflies affects tadpole shape and color. Oecologia 109: 615–621. McCollum SA, Van Buskirk J. 1996. Costs and benefits of a predator-induced polyphenism in the gray treefrog Hyla chrysoscelis. Evolution 50: 583–593. Milliken GA, Johnson DE. 1984. Analysis of Messy Data. New York: Van Nostrand Reinhold. Moore RD, Newton B, Sih A. 1996. Delayed hatching as a response of streamside salamander eggs to chemical cues from predatory sunfish. Oikos 77: 331–335. Newman RA. 1992. Adaptive plasticity in amphibian metamorphosis. Bioscience 42: 671–678. Nodzenski E, Wassersug RJ, Inger RF. 1989. Developmental differences in visceral morphology of megophryine pelobatid tadpoles in relation to their body form and mode of life. Biological Journal of the Linnean Society 38: 369–388.

HATCHING AGE, MORPHOLOGY AND DEVELOPMENT

469

Parichy DM, Kaplan RH. 1992. Maternal effects on offspring growth and development depend on environmental quality in the frog Bombina orientalis. Oecologia 91: 579–586. Parichy DM, Kaplan RH. 1995. Maternal investment and developmental plasticity: functional consequences for locomotor performance of hatchling frog larvae. Functional Ecology 9: 606–617. Petranka JW, Just JJ, Crawford EC. 1982. Hatching of amphibian embryos: the physiological trigger. Science 217: 257–259. Petranka JW, Sih A, Kats LB, Holomuzki JR. 1987. Stream drift, size-specific predation and the evolution of ovum size in an amphibian. Oecologia 71: 624–630. Rowe L, Ludwig D. 1991. Size and timing of metamorphosis in complex life cycles: time constraints and variation. Ecology 72: 413–427. Seymour RS, Bradford DF. 1995. Respiration of amphibian eggs. Physiological Zoology 68: 1–25. Seymour RS, Geiser F, Bradford DF. 1991. Gas conductance of the jelly capsule of terrestrial frog eggs correlates with embryonic stage, not metabolic demand or ambient PO2. Physiological Zoology 64: 673–687. Seymour RS, Mahony MJ, Knowles R. 1995. Respiration of embryos and larvae of the terrestrially breeding frog, Kyarranus loveridgei. Herpetologica 51: 369–376. Siegel S, Castellan NJ, Jr. 1988. Non-parametric statistics for the behavioral sciences. New York: McGrawHill. Sih A, Moore RD. 1993. Delayed hatching of salamander eggs in response to enhanced larval predation risk. American Naturalist 142: 947–960. Skelly DK, Werner EE. 1990. Behavioral and life-historical responses of larval american toads to an odonate predator. Ecology 71: 2313–2322. Smith DC, Van Buskirk J. 1995. Phenotypic design, plasticity, and ecological performance in two tadpole species. American Naturalist 145: 211–233. Sokal RR, Rohlf FJ. 1995. Biometry: the principles and practice of statistics in biological research. New York: W. H. Freeman. Travis J. 1994. Evaluating the adaptive role of morphological plasticity. In: Wainwright, PC, Reilly, SM, eds. Ecological morphology: integrative organismal biology. Chicago: University of Chicago Press, 99–122. Viertel B. 1985. The filter apparatus of Rana temporaria and Bufo bufo larvae (Amphibia, Anura). Zoomorphology 105: 345–355. Viertel B. 1989. The filter apparatus of anuran larvae – aspects of the filtering mechanism. In: Splechtna, H, Hilgers, H, eds. Trends in vertebrate morphology. New York: Gustav Fischer Verlag, 526–533. Viertel B. 1991. The ontogeny of the filter apparatus of anuran larvae (Amphibia, Anura). Zoomorphology 110: 239–266. Warkentin KM. 1995. Adaptive plasticity in hatching age: a response to predation risk trade-offs. Proceedings of the National Academy of Sciences 92: 3507–3510. Warkentin KM. 1998. Phenotypic plasticity at hatching in the red-eyed treefrog, Agalychnis callidryas: life history, behavior and development. Unpublished D. Phil. Thesis, University of Texas at Austin. Warkentin KM. 1999. The development of behavioral defenses: a mechanistic analysis of vulnerability in red-eyed tree frog hatchlings. Behavioral Ecology 10: 251–262. Wassersug R. 1972. The mechanism of ultraplanktonic entrapment in anuran larvae. Journal of Morphology 137: 279–288. Wassersug RJ. 1976. Oral morphology of anuran larvae: terminology and general description. Occasional Papers of the Museum of Natural History, The University of Kansas 48: 1–23. Wassersug RJ. 1980. Internal oral features of larvae from eight anuran families: functional, systematic, evolutionary and ecological considerations. University of Kansas Museum of Natural History, Miscellaneous Publication 68: 1–146. Werner EE. 1986. Amphibian metamorphosis: growth rate, predation risk, and the optimal size at transformation. American Naturalist 128: 319–341. Werner EE, Gilliam JF. 1984. The ontogenetic niche and species interactions in size structured populations. Annual Review of Ecology and Systematics 15: 393–425. Werner EE, Gilliam JF, Hall DJ, Mittelbach GG. 1983a. An experimental test of the effects of predation risk on habitat use in fish. Ecology 64: 1540–1548. Werner EE, Hall DJ. 1988. Ontogenetic habitat shifts in bluegill: the foraging rate–predation risk trade-off. Ecology 69: 1352–1366.

470

K. M. WARKENTIN

Werner EE, Mittelbach GG, Hall DJ, Gilliam JF. 1983b. Experimental tests of optimal habitat use in fish: the role of relative habitat profitability. Ecology 64: 1525–1539. Winer BJ. 1971. Statistical principles in experimental design. New York: McGraw-Hill. Wourms JP. 1972. The developmental biology of annual fishes: III. Pre-embryonic and embryonic diapause of variable duration in the eggs of annual fishes. Journal of Experimental Zoology 182: 389–414.