Glassfrog embryos hatch early after parental desertion

Jesse R. J. Delia, Aurelio Ramírez-Bautista, Kyle Summers


Both parental care and hatching plasticity can improve embryo survival. Research has found that parents can alter hatching time owing to a direct effect of care on embryogenesis or via forms of care that cue the hatching process. Because parental care alters conditions critical for offspring development, hatching plasticity could allow embryos to exploit variation in parental behaviour. However, this interaction of parental care and hatching plasticity remains largely unexplored. We tested the hypothesis that embryos hatch early to cope with paternal abandonment in the glassfrog Hyalinobatrachium fleischmanni (Centrolenidae). We conducted male-removal experiments in a wild population, and examined embryos' response to conditions with and without fathers. Embryos hatched early when abandoned, but extended development in the egg stage when fathers continued care. Paternal care had no effect on developmental rate. Rather, hatching plasticity was due to embryos actively hatching at different developmental stages, probably in response to deteriorating conditions without fathers. Our experimental results are supported by a significant correlation between the natural timing of abandonment and hatching in an unmanipulated population. This study demonstrates that embryos can respond to conditions resulting from parental abandonment, and provides insights into how variation in care can affect selection on egg-stage adaptations.

1. Introduction

The early stages of life are often the most vulnerable, and developing embryos in eggs have traditionally been viewed as helpless. Parental care of eggs has evolved many times, presumably because of the advantages of protecting vulnerable young [1]. Nonetheless, embryos of many species are capable of self-defence by hatching early to escape danger or by delaying hatching when eggs are safe (reviewed by Warkentin [2]). Hatching plasticity encompasses a diversity of phenomena, from environmental effects on embryogenesis that alter the timing of hatching to active embryo responses that improve fitness in dynamic environments [3]. It is likely that parental care affects the evolution of hatching plasticity in important ways. However, how embryo strategies and parental behaviour interact—in individual lives or through evolutionary processes—remains relatively unexplored.

Warkentin [2,3] outlined two types of developmental changes that can lead to environmentally cued plasticity in hatching time. When hatching occurs at some fixed point in development, hatching plasticity may occur by altering the rate of embryonic development (i.e. accelerations or decelerations). This can result from an inevitable effect of the environment on physiology (e.g. temperature affects metabolism and development rate) as well as sophisticated mechanisms that allow environmentally cued flexibility in development (e.g. diapause). When the rate of development remains consistent, hatching plasticity can occur by altering the developmental sequence (i.e. hatching earlier or later relative to other developmental events). Both mechanisms may involve traits that enable survival in or outside the egg at different stages of development (e.g. behaviour, physiology, maternal yolk provisions) as well as those that permit environmentally cued hatching (e.g. sensory systems, hatching mechanisms). Parental care can have both direct and indirect effects on these developmental processes [25].

Available information indicates that hatching plasticity can result from a direct effect of parental care on embryogenesis or via forms of care that benefit offspring [25]. For example, parental incubation of bird and reptile eggs accelerates development owing to temperature effects on embryogenesis [6,7]. Variation in incubation behaviour can have important consequences for both parental investment and offspring development [4,7,8]. In fact, the variable outcomes of parental care on embryogenesis have been documented across the animal kingdom, demonstrating that parental traits can directly impact phenotypic development [5,6,9,10].

Parents can also cue and/or assist the hatching process—termed ‘hatching care’ by Mukai et al. [11]. Mother shield-bugs vibrate their eggs to initiate hatching, and removing mothers reduces hatching rates [11]. Father mudskippers actively flood egg chambers to cue hatching at high tides [12]. In some species, parents can alter hatching time by cueing changes in developmental rate, such as maternal control of egg diapause in insects [5]. Embryos are also known to signal parents when they are ready to hatch. For example, embryonic vocalizations stimulate mother crocodiles to open nests [13]. The hatching process of many decapod crustaceans is activated by abdominal pumping of egg-carrying parents, which synchronizes larval release with tidal cycles; in some species, this process involves feedbacks between parental behaviour and offspring chemical cues [14]. Among known examples, parentally cued hatching appears to be adaptive to offspring survival, occurring in response to optimal conditions for hatchlings [2,5,12,14] and predators [15], and/or assisting the hatching process [11,13]. Nonetheless, many studies have not evaluated parental traits within a cost–benefit framework (but see [12]), and fewer have tested whether parents cue hatching owing to constraints on providing care.

Because parental care alters conditions necessary for offspring development, hatching plasticity could allow embryos to exploit variation in parental behaviour. Parental care in many oviparous animals functions to buffer the egg stage from variation in environmental conditions (e.g. temperature, hydration, oxygen and predation [1]). This buffering effect should relax selection on the suites of traits that permit environmentally cued hatching and potentially lead to a fixed embryonic period via a process of canalization [1618]. However, variation in parental behaviour occurs throughout the animal kingdom, often resulting from an inherent trade-off in parental investment [1]. A history of variable parental care may generate environmental instability during the egg stage, altering the optimal time to hatch, and thereby expose embryos to selection on plasticity in hatching time. Under such conditions, embryos could improve survival by hatching in response to cues that indicate parental abandonment.

While research to date has found that parents can alter hatching time via direct effects on embryogenesis and/or by cueing hatching, the hypothesis that embryos time hatching to cope with variation in parental care remains largely unexplored. In the frog Eleutherodactylus coqui, males guard embryos from predators, but frequently abandon nests or eat their eggs, resulting in a variable egg environment [19,20]. Buckley et al. [20] induced early hatching in this species by mimicking predation in the laboratory, suggesting that embryos may respond to factors associated with conditional paternal care in the wild. However, a key question that remains is: can embryos time hatching to exploit variation in parental care?

Neotropical glassfrogs (Centrolenidae) offer excellent opportunities to investigate parent–embryo interactions. Like other frogs with arboreal embryonic development and aquatic larvae, hatching in glassfrogs involves a dramatic habitat shift that provides opportunities for stage-specific selection [21]. In contrast to other such species where hatching plasticity has been studied [21,22], many glassfrogs provide parental care to their eggs [23]. In previous research [24], we examined the environmental contingency of male care in Hyalinobatrachium fleischmanni (figure 1). In this species, fathers hydrate eggs by performing brooding behaviour, which functions to prevent embryonic dehydration [24]. Paternal hydration is the primary source of moisture for embryos, because oviposition sites on the lower surfaces of leaves shield egg clutches from rainfall [24]. Fathers initiate clutch hydration and maintain it during development by adjusting egg-brooding frequency in response to dehydration and weather conditions [24]. The duration of male care varies from 3 to 19 days after oviposition, and there is an extensive plastic-hatching period ranging from 12 to 27 days past oviposition (figure 1). We found that mating opportunities can conflict with paternal behaviour and increase mortality for younger embryos (less than 72 h old), when males temporarily abandon clutches for up to 48 h during courtship [24]. Hence, male responses to social and abiotic factors can provide a source of environmental heterogeneity for eggs. Considering that paternal behaviour is critically tied to egg–water balance, it is possible that hatching plasticity allows embryos to improve survival when they are abandoned.

Figure 1.

Summary of embryonic development, hatching plasticity and paternal care in H. fleischmanni. Development: filled circles and solid line show mean time from oviposition until marker developmental stages [24,25]; error bars are standard deviation across clutches (n = 36). The last data point indicates the latest hatching observed. No embryos reached stage 26 before hatching, but substantial development occurred. Inset: ventral view of hatchlings preserved at the average onset of hatching competence (12 days) and the average termination of hatching when receiving care (22 days). Note the intact yolk sac in the early hatchling versus functional digestive system in the older animal, as well as the development of branchial baskets and keratinization of jaw sheaths/denticles. Care and hatching: dashed line shows the range of clutch ages at which unmanipulated males terminated care (open square is mean, n = 18). Dotted line shows the range of ages at which embryos hatched (open circle is mean, n = 36). Shaded rectangles show periods of obligate (stippled) and facultative (grey) parental care. Male removal within 48 h of oviposition resulted in 90% embryo mortality, but decreased to 7% for removals occurring at or after 5 days [24]. (Online version in colour).

We tested the hypothesis that H. fleischmanni embryos hatch early when abandoned. We conducted a male-removal experiment and examined embryos' response to egg-stage conditions with and without fathers. Distinguishing an embryo response from a direct effect of care on developmental rate requires an assessment of the timing of hatching, the rate of embryonic development and the phenotypes of hatchlings [2]. We compared the developmental sequence of hatching across treatments to determine whether care directly affects developmental rate or embryos actively hatch at different developmental stages. To corroborate our experimental results, we examined the natural timing of paternal abandonment and hatching in an unmanipulated population. We discuss the implications of our results with respect to variability in care as a factor favouring adaptive embryo strategies.

2. Methods

We investigated the interaction of paternal care and hatching plasticity in a population of H. fleischmanni near San Gabriel Mixtepec in Oaxaca, Mexico, during the rainy seasons of 2009 and 2010. Male territories were monitored nightly along stream transects to observe mating activity and to record newly deposited clutches. These clutches were monitored daily throughout development until embryos hatched or died. Specific types of embryonic mortality were identified using established methods [24,26]. Hatching cups were used to determine the onset and conclusion of hatching for each clutch [26]. Embryonic development was staged daily according to Gosner [25].

(a) Effect of paternal abandonment on hatching

To test the hypothesis that embryos hatch early when abandoned, we conducted a late-removal experiment by displacing 40 males from their clutches between 2 and 8 days after oviposition. Embryo survival, development and hatching time were compared with the clutches of 50 attending males; clutch size was not different between treatments (t-test; t82 = −1.51, p = 0.134). Rainfall was measured daily to evaluate whether embryos respond to variation in weather conditions that increase the rate of clutch dehydration [24]. To corroborate our experimental results, we investigated whether the timing of hatching was affected by the natural duration of paternal care in an unmanipulated group of 18 males. Paternal abandonment was determined by monitoring clutches during 8 h periods on four of every five nights from oviposition until hatching.

Separate independent-sample t-tests were used to compare the number of days until the onset and until the conclusion of hatching between treatments; Welch's t-test was used when unequal variances were detected. Thirty-two removal and 45 control clutches survived until hatching; 10 control males (20%) abandoned their clutch 5–9 days prior to the onset of hatching competence, five of which survived. To compare the effect of male removal with clutches receiving care, we excluded these five abandoned clutches from initial analyses, but included them in analyses of natural abandonment (see below). Mann–Whitney U-tests were used to examine survival rates between treatments by comparing the proportion of mortality among individual clutches. To determine whether embryos respond to changes in environmental conditions without fathers, we modelled the influence of rainfall on the timing of hatching using ANCOVAs. For each clutch, rainfall totals were calculated from oviposition until hatching competence (stage 25, Embedded Image = day 12.24), and entered as a covariate in the model. For the unmanipulated population, we used linear regression to examine the relationship between the timing of paternal abandonment (measured as the last day a male brooded his clutch) and the conclusion of hatching for each clutch.

(b) Effect of paternal care on embryonic development and hatchling phenotypes

At this site, H. fleischmanni embryos can remain in the egg for 16 days after reaching hatching competence occurring at stage 25 (figure 1). However, established staging tables do not offer sufficient resolution to monitor the substantial morphological changes that occur during this period (i.e. organ development), owing to a several-month period between stage 25 and stage 26. To examine the effects of paternal care on changes in developmental rate and hatchling phenotypes, we performed morphological analyses on hatchlings collected during the removal experiment. Hatchlings from seven removal and 10 control clutches were not measured owing to distorted morphology (rainfall diluted osmolytes in hatching cups). Specimens were photographed using a Sony HD handycam attached to an Olympus SZX9 dissecting microscope. Image analyses were conducted using ImageJ v. 1.46 (NIH). We measured growth as total length (TL, mm) and development by counting the number of gut-coil rotations. These were averaged for each clutch, then log-transformed for analyses.

To test the hypothesis that hatching plasticity is due to embryos hatching at different stages of development, rather than a direct effect of care on developmental rate, we used ANCOVAs to compare both phenotypic parameters separately. Age of hatching was added as a covariate in the model; an age-by-treatment interaction on phenotypic parameters would indicate differential rates of development and support the hypothesis that hatching plasticity is due to direct effects [2]. The distribution of gut-coil development was quasi-binomial in removals, whereas controls were normal; removal embryos hatched earlier with few or no gut coils, whereas controls hatched later with one or more gut coils. To account for this, we first tested the treatment effect on gut-coil development for all samples using a Mann–Whitney U-test, then restricted the ANCOVA to similar-aged treatments; this normalized the residuals in the removal group (Kolmogorov–Smirnov test, p = 0.168).

3. Results

(a) Effect of paternal abandonment on hatching

Removing fathers significantly reduced the amount of time until the onset and conclusion of hatching (t70 = 5.84, p < 0.0001 and Welch's t46.06 = 6.45, p < 0.0001, respectively; figure 2). Males among the unmanipulated population terminated care from 3 to 19 days past oviposition (12.11 ± 4.55 [mean ± s.d.]). The natural timing of paternal abandonment had a strong positive effect on the length of the embryonic period (linear regression: r2 = 0.696, F1,16 = 36.62, p < 0.0001, n = 18; figure 3). Embryo survival did not differ between treatments, whether comparing all types of mortality, or specifically mortality owing to dehydration (Mann–Whitney U = 1064.50, p = 0.060, n = 84 and U = 947.0, p = 0.250, respectively).

Figure 2.

Effect of male removal on the hatching age of their clutches. (a,c) Box and whisker plots show mean, 95% CI, and outliers for removal clutches and control clutches receiving care (***p < 0.0001). (b,d) Scatter plots show the relationship of rainfall to hatching timing across treatments. Embryonic periods increased with rainfall; however, the main effect of care was significant after accounting for variation attributed to rainfall (table 1).

Figure 3.

Natural covariation between the timing of paternal abandonment and the conclusion of hatching.

On average, there was a 21.2% reduction in the duration of the embryonic period for the removal treatment, although differences ranged up to 34.6% during drier periods. Rainfall had a positive effect on the embryonic period in both treatments (table 1 and figure 2). For the onset of hatching, there was no difference in the strength of this interaction between treatments. However, for the conclusion of hatching, rainfall had a stronger effect on the removal treatment. Nonetheless, hatching was still significantly earlier in the removal treatment for both parameters even at the highest rainfall levels (table 1 and figure 2).

View this table:
Table 1.

ANCOVA for effects of rainfall, paternal care treatment and their interaction on hatching age. Sum of squares (ss) are type III.

Early hatching appeared to occur in response to deteriorating conditions without fathers. Removal clutches lost thickness (a measure of hydration) over time, and were significantly thinner than control clutches at the onset of hatching competence (t43.0 = 9.94, p < 0.0001, n = 45). Embryos from both treatments hatched several days after abandonment (figures 2 and 3); removals occurred 4–10 days prior to stage 25 and hatching often began 0–48 h after reaching stage 25 (59.38% of clutches). Paternal care did not have an obvious effect on egg integrity (i.e. egg capsules did not degrade over time), as egg clutches remained conspicuous and resilient structures even two weeks after hatching. During field monitoring, we anecdotally observed hatching in four removal and five control clutches—in all cases, embryos used physical exertion during hatching, wiggling to exit the egg capsule.

(b) Effect of paternal care on embryonic development and hatchling phenotypes

Variation in the embryonic period spanned 16 days for both the onset and the conclusion of hatching across all clutches. This represents a magnitude of 59% hatching plasticity considering the maximum embryonic period or 49.6% considering the average among control clutches. Within clutches, development was synchronous for all daily checks except during gill regression (stages 23–25), during which time some embryos resorbed the first gill filament (stage 24) or completed gill regression (stage 25) up to 24 h earlier than their clutch mates. Most embryos hatched in stage 25, but a few from three removal clutches hatched in stages 23 and 24 with similar-aged stage 25 clutch-mates. Embryos reached stage 25 between days 11 and 14 (12.24 ± 0.14).

Removing fathers had no effect on the rate of embryonic development to stage 25 (Welch's t27.92 = −0.109, p = 0.914, n = 45). Hatchlings from removal clutches were significantly smaller, with significantly fewer gut coils than control clutches (Welch's t27.97 = 5.42, p < 0.0001 and Mann–Whitney U: U = 187.0, p = 0.004, n = 54, respectively). The results of the ANCOVAs reject the hypothesis that differences in hatching time were due to parental effects on the rate of embryonic development (table 2). Comparisons of TL revealed that embryos grew continuously and steadily in length during development, and this measure of growth rate did not differ between treatments (table 2a). The results for gut-coil development among similar-aged clutches found that age had the only significant effect on development (table 2b). Therefore, phenotypic differences between treatments resulted from embryos hatching at different stages of development. Dissections of three control embryos that hatched near the average hatching time found that they had functioning gut coils, indicated by the presence of a lumen and faecal production (they presumably ate immediately following hatching when restrained in cups). By contrast, removal hatchlings had fewer gut coils that were filled with yolk, indicating that differences in hatching age between treatments reflect the onset of feeding competence (figure 1).

View this table:
Table 2.

ANCOVA for effects of paternal care treatment, developmental rate (age) and their interaction on hatchling phenotype. Sum of squares (ss) are type III.

4. Discussion

We tested the hypothesis that H. fleischmanni embryos hatch early when abandoned. Removing fathers resulted in embryos hatching early in development, whereas control embryos extended development under the safety of care (figure 2 and table 1). Among an unmanipulated population, the natural timing of paternal abandonment explained 70% of the variation in the duration of the egg stage (figure 3). Paternal care had no direct effect on the rate of embryonic development. Rather, hatching plasticity was due to embryos actively hatching at different developmental stages, probably in response to deteriorating conditions without fathers (table 2 and figure 1). To the best of our knowledge, this is the first study demonstrating that embryos time hatching to cope with variation in parental care.

Research has found that parents can alter hatching time due to a direct effect of care on embryonic development or via forms of care that benefit offspring [215]. By contrast, our study supports the idea that variation in parental care generates conditions that favour adaptive embryo strategies. Hyalinobatrachium fleischmanni fathers did not alter development rate; nor did we find evidence that they cued hatching. Instead, our results indicate that embryos are responsive to deteriorating egg environments in the absence of care. Dehydration-induced hatching has been documented in two species of arboreal-breeding frogs that lack parental care [21,22]. In H. fleischmanni, fathers are the primary source of hydration for embryos since oviposition sites shield clutches from rainfall [24]. Differences in egg hydration levels between treatments clearly indicate that paternal behaviour is a key source of environmental heterogeneity for embryos. Although rainfall can slow the rate of egg water loss and interact with the duration of embryonic development, eggs only dehydrate when they are abandoned (which occurs from 3 to 19 days after oviposition) [24]. It is likely that the use of dehydration as a proximate mechanism would allow embryos to respond to abandonment and exploit good conditions when fathers continue to provide care. However, additional research needs to examine cue use by H. fleischmanni embryos.

To understand the adaptive value of plasticity requires information on both the costs and the benefits that characterize selective trade-offs [27]. It is clear that early hatching improves egg-stage survival when fathers abandon their embryos. We conducted a late-removal treatment and found no effect on embryonic mortality—this was probably due to the hatching abilities of embryos, rather than a decline in the importance of care per se. Whether delayed hatching can improve offspring survival remains to be tested. However, we did find that removal hatchlings were smaller and less developed than controls, which may carry a performance cost in the larval stage [2,21]. In addition, dissections revealed that control embryos hatched at or near feeding competence, whereas removal embryos hatched with intact yolk sacs. In many terrestrially breeding frogs with aquatic larvae, tadpoles hatch prior to feeding competence and use yolk reserves for initial growth during the larval stage [21,22]. Continued care probably allows H. fleischmanni embryos to delay hatching in order to maximize development using yolk and to be feeding-competent at hatching, thereby minimizing lag-time in exotrophically based growth [16,28].

Variation in parental investment is thought to influence the evolution of offspring adaptations [29]. Research on hatching plasticity offers new prospects to examine parent–offspring interactions among species where parental care is limited to the egg stage. In many oviparous animals, parental behaviour modulates both the biotic and abiotic environments of eggs [1], and social dynamics of parents can produce variable conditions for embryos [30,31]. In H. fleischmanni, we previously found that mating opportunities can conflict with paternal care and increase embryo mortality during dry periods [24]. The finding that paternal care and hatching plasticity are critically connected in ecological time suggests that embryo hatching may be evolving in response to variation in parental investment in this system. Research in other animals could reveal general dynamics of parent–embryo coevolution, and offer opportunities to address how social factors influence egg-stage adaptations.

5. Conclusion

Environmentally cued hatching has been documented across the animal kingdom, indicating that some form of hatching plasticity is likely to be an ancestral condition [2]. What is not known is how parental care affects the evolution of hatching plasticity. By reducing egg-stage mortality, the evolution of parental care is predicted to favour larger eggs and longer embryonic periods [16,17]. While such a process should relax selection on hatching plasticity, our study offers insights into how variation in care may expose embryos to egg- and larval-stage selection. We argue that this interaction may be more general than is currently appreciated and encourage future investigations in other species. Integrative and comparative research among other groups will contribute to our understanding of how parental care influences the evolution of embryo strategies.

Funding statement

Funding was provided by grant-in-aid of research from Sigma Xi, the American Society of Ichthyologists and Herpetologists Gaige Fund Award, and the Next-Step Scholarship from East Carolina University. Permits were provided by the Secretaria de Medio Ambiente y Recursos Naturals (nos. 01902, 01903, 09279, 09280).

Data accessibility

The data relating to this study are available in the electronic supplementary material.


We acknowledge all the thoughtful comments, suggestions and critical reviews of methods, analyses and/or versions of this manuscript by K. M. Warkentin, R. W. McDiarmid, D. R. Chalcraft, S. McRae, J. Tumulty, K. L. Cohen and J. C. Touchon. The associate editor and two anonymous reviewers provided insightful and valuable comments that greatly improved the quality of the manuscript. M. Trez assisted with measuring specimens.

  • Received December 11, 2013.
  • Accepted March 26, 2014.


View Abstract