Parent–offspring similarity in the timing of developmental events: an origin of heterochrony?

Oliver Tills, Simon D. Rundle, John I. Spicer

Abstract

Understanding the link between ontogeny (development) and phylogeny (evolution) remains a key aim of biology. Heterochrony, the altered timing of developmental events between ancestors and descendants, could be such a link although the processes responsible for producing heterochrony, widely viewed as an interspecific phenomenon, are still unclear. However, intraspecific variation in developmental event timing, if heritable, could provide the raw material from which heterochronies originate. To date, however, heritable developmental event timing has not been demonstrated, although recent work did suggest a genetic basis for intraspecific differences in event timing in the embryonic development of the pond snail, Radix balthica. Consequently, here we used high-resolution (temporal and spatial) imaging of the entire embryonic development of R. balthica to perform a parent–offspring comparison of the timing of twelve, physiological and morphological developmental events. Between-parent differences in the timing of all events were good predictors of such timing differences between their offspring, and heritability was demonstrated for two of these events (foot attachment and crawling). Such heritable intraspecific variation in developmental event timing could be the raw material for speciation events, providing a fundamental link between ontogeny and phylogeny, via heterochrony.

1. Introduction

How ontogeny (development) and phylogeny (evolution) are linked has been, and remains, one of the key questions in biology. In fact, heterochrony, the altered timing of developmental events between ancestors and descendants, has been suggested as the main driver of evolutionary change [1,2]—a suggestion reinforced by the fact that heterochrony has been documented as an evolutionary pattern in several animal groups and for a diverse array of morphological [36] and physiological [7,8] traits. Despite the pervasiveness of the notion that heterochrony has a genetic basis, we still know relatively little about the processes through which this pattern forms. An obvious candidate process is that intraspecific variation in developmental event timing provides raw material on which selection, either during or after development, could act [9]. However, heterochrony has typically been studied using between-species comparisons, with an assumption that intraspecific developmental event timing is largely invariant or insignificant, despite variation being required for evolutionary change. Consequently, we know far less about intraspecific differences in developmental event timing, even though variation at this level is a probable source of answers to questions regarding the formation of heterochronies.

Recently, an increasing number of studies have reported widespread intra-specific variation in developmental event timing [1016]. For example, de Jong [15] showed, using 82 developmental characters in 261 embryos of the cichlid, Haplochromis piceatus, that embryos could follow one of 26 880 different developmental sequences [15]. This magnitude (and complexity) of intraspecific variation is astounding and demonstrates that intraspecific developmental event timing is far from invariant.

While intraspecific variation in developmental event timing clearly exists, its source and relationship to between-species differences in developmental event timing are not clear. Tills et al. [17] collected embryos of the pond snail Radix balthica from a single population and compared the degree of inter-individual difference in developmental event timing with pairwise genetic distance. The magnitude and number of inter-individual developmental event timing differences increased with increasing genetic distance, providing support for there being a genetic basis for variation in the timing of these embryonic developmental events. For intra-specific variation in developmental event timing to be the raw material from which heterochronies arise requires that it should have both a genetic basis and be heritable. Heritability of developmental event timing would point to intraspecific variation, of which there is increasing evidence [10,15], as an origin of the macroevolutionary pattern of heterochrony.

Here, we used R. balthica to perform a direct parent–offspring comparison to investigate whether cross-generational similarity in developmental event timing was evident, and whether any such similarity extends to the level of heritability. A limiting factor in any study of developmental event timing is the resolution (both temporal and spatial) with which organisms can be observed. Consequently, we used a custom bio-imaging system developed in our laboratory for the specific task of monitoring large numbers of aquatic embryos with high temporal and spatial resolution. Parents were harvested as embryos from a stock population and this system was used to video their entire embryonic development (20 s recording made every 1 h for parents and every 2 h for offspring). Once hatched, these snails were cultured in isolation through to reproduction and, as a result of R. balthica being a simultaneous hermaphrodite, offspring were produced sexually [18,19]. The embryonic development of these offspring was then recorded exactly as for their parents. Video for both parents and offspring was then analysed to determine, with high precision, the time of occurrence of 12 key developmental events. This is a level of precision that is beyond what has been achieved previously in such developmental studies. A limitation of direct parent–offspring comparisons is the inability to control for maternal effects [20] and therefore here we used egg size as a measure of maternal investment [2123] and incorporated this measure into analyses to estimate its contribution to the observed relationships and to investigate relationships with its effect removed.

2. Material and methods

(a) Animal culture

To begin the parent–offspring comparison, embryos were harvested from an F2 laboratory stock population of Radix balthica originating from the River Dart, Totnes, Devon, UK (50°26′19′ N, 3°41′24′ W). Embryos were examined under low power magnification (×10) and those that had not developed past the two-cell division stage were dissected from their egg masses and placed individually within a microtitre plate (384 well plate, volume (vol.) per well = 70 μl) containing artificial pond water (APW) [24]. This microtitre plate was placed into a custom system, designed for time-lapse inverted imaging of aquatic embryos, which is both described below and depicted in figure 1. The plate was subject to continuous illumination and housed in a controlled temperature facility (T = 20°C). Water changes were performed every two days and each well was checked to identify hatchlings daily.

Figure 1.

Schematic of the custom-built bio-imaging system used in this study. Further details of the specific components used are given in §2. (Online version in colour.)

Hatchlings were cultured in glass jars (vol. = 40 ml) containing APW (vol. = 35 ml) and a 5 cm length of the pond weed, Elodea densa. Jars were placed on shelving positioned in a west-facing window (in locations chosen at random and changed every 21 days) in a 15°C controlled temperature facility. Snails were provided with washed lettuce discs (diam. = 4 mm) every two weeks. Mortality in the three months following hatching was 50%, reducing the number of snails cultured from 28 to 14. Three months after hatching, snails were transferred to larger jars (vol. = 500 ml) containing APW (vol. = 450 ml) and two 15 cm lengths of Elodea. Water was changed in these jars every 21 days and snails were fed discs of washed lettuce (diam. = 10 mm) every 14 days. When snails reached maturity, jars were checked regularly for egg masses. These masses were examined under low- power magnification and, if embryos had not developed past the two-cell division stage, their entire embryonic development was imaged using the same method as for their parents' embryonic development. Ten out of the 14 snails that grew to a mature size reached sexual maturity and produced viable eggs; however, one of these snails produced only a single egg and therefore this individual is not included in the analyses presented here. The nine remaining snails produced between three and 19 viable embryos.

(b) Image acquisition

A 150-frame image sequence of each embryo was acquired (at 7.5 frames s–1) every 2 h until hatching, using a custom-built imaging system comprising an aluminium frame housing an Optiscan (Prior Scientific, UK) XY motorized stage mounted above a Pike 421B (Allied Vision Technology, Statdtroda, Germany) four MP monochrome camera connected to a VHZ20R (Keyence, Milton Keynes, UK) zooming lens with dark-field illumination provided by an LED array (figure 1). The motorized stage and camera were controlled and synchronized using the open source software Micromanager v. 1.3 [25] run on a Mac Pro.

(c) Image analysis

From the second cell division onwards, the time of onset of 12 developmental events (table 1) was determined by manual observation of the image sequences recorded for both parental and offspring embryonic development. The developmental events used here can be categorized according to whether they can be identified using a still image (velum, trochophore, veliger, eye spot formation, shell formation, capsule rupture, hatching) or those that require video (heart beating, body flexing, foot attachment, crawling, use of radula). Figure 2 presents the developmental events identifiable using still images, including features both before and after each event. The specific nature of the events requiring video can be viewed in the electronic supplementary material.

View this table:
Table 1.

Descriptions of the embryonic developmental events used (based on [26,28]).

Figure 2.

Images of Radix balthica at the specific developmental events used here and that are identifiable using still images: velum—(a)—the semi-transparent velum has extended until it fully covers one side of the embryo; trochophore—(b)—circular liver cells are visible in the embryo and have extended to be apparent on both sides; veliger—(c)—the two velar lobes have extended, forming a ‘crescent moon’ shape with a clear indentation being visible between the lobes; eye spot formation—(d)—both eye spots are evident as light grey spots on either side of the head; shell formation—(e)—the shell ridge becomes visible on the top of the mantle as an elevation that, as development progresses, grows forward and down towards the head; capsule rupture—(f)—the smooth spheroid surface of the egg is disturbed, instead appearing crumpled, once the embryo punctures the capsule using its radula.

Egg length and width were measured from the images at the time of four-cell division. From these values, egg volume was calculated using the following formula for a regular prolate spheroid shape [27]:Embedded Image where l = egg length and w = egg width. Egg volume was subsequently incorporated into our analysis to investigate its contribution towards observed relationships [20].

3. Results

To test for differences in developmental event timing between the offspring from different parents and for a relationship in event timing between parents and their offspring, we performed an ANOVA with parents weighted by their own event timing and offspring egg volume included as a covariate (figures 3 and 4; table 2). The timings of all developmental events, with the exception of capsule rupture, were significantly different between the offspring from different parents. Predicted values for offspring timing of each developmental event, produced by the ANOVA, were related positively with parental timing, demonstrating parent–offspring similarity in the timing of all 12 of the developmental events used here (figures 3 and 4).

Figure 3.

(af) Box plots (median ± first/third quartile; dotted lines indicate the full data range) showing the timing of the first six developmental events used here, for offspring produced from different parents (indicated by different colours/shades) arranged in increasing order of their event timing. Significant differences in developmental event timing between offspring produced from different parents are indicated by horizontal bars above box plots (determined by ANOVA between parents weighted with parental developmental timing using weighted least squares and with offspring egg volume included as a covariate). Plots of developmental event timing (mean ± 1 s.e.) for offspring produced from different parents, predicted from the ANOVA model, are shown for each developmental event above the box plots. (Online version in colour.)

Figure 4.

(af) Box plots (median ± first/third quartile; dotted lines indicate the full data range) showing the timing of the last six developmental events used here, for offspring produced from different parents (indicated by different colours/shades) arranged in increasing order of their event timing. Significant differences in developmental event timing between offspring produced from different parents are indicated by horizontal bars above box plots (determined by ANOVA between parents weighted with parental developmental timing using weighted least squares and with offspring egg volume included as a covariate). Plots of developmental event timing (mean ± 1 s.e.) for offspring produced from different parents, predicted from the ANOVA model, are shown for each developmental event above the box plots. (Online version in colour.)

View this table:
Table 2.

Results of an ANOVA testing for differences in the timing of developmental events (see electronic supplementary material, supplementary table 1) between offspring produced from different parents with parents weighted by their own developmental timing, using weighted least squares and offspring egg volume included as a covariate.

Significant heritabilities of event timings, calculated by regressing the average timing of each event in offspring on that in their parents [20], were found for foot attachment (h2 = 0.380) and crawling (h2 = 0.381) (figure 5). An obvious limitation of direct parent–offspring comparisons is the inability to control for maternal effects. To assess whether relationships between parents and their offspring in the timing of foot attachment and crawling were being driven by differences in egg volume, we performed the same analysis, but replaced means for offspring developmental event timing with means of the residuals from regressions, testing for the effect of egg volume on the timing of these developmental events (figure 5). This procedure effectively removes any effect of egg volume on offspring developmental timing. This analysis resulted in no significant change to the relationships between parents and their offspring in the timing of foot attachment and crawling, and again demonstrated heritability for these events, indicating that egg volume was not a driver of these relationships. In fact, the relationship between the timing of shell formation in parents and their offspring was only significant when the effect of egg volume was removed, suggesting that egg volume was actually masking the relationship indicative of heritability in the timing of shell formation (figure 5).

Figure 5.

Comparison of parental event timing with: (bottom plot) mean (± 1 s.e.) offspring timing, to test for heritability (foot attachment—R2 = 0.47, F1,7 = 6.28, p = 0.041; regression coefficient (h2) = 0.380. Crawling—R2 = 0.44, F1,7 = 6.22, p = 0.041; regression coefficient (h2) = 0.381), (top plot) mean (± 1 s.e.) offspring residuals, from a regression analysis testing for the effect of egg volume on event timing (blue/grey plots in top panel). To examine whether relationships that indicate heritability are still present with the effect of egg volume on developmental timing removed. (a) Shell formation—R2 = 0.39, F1,7 = 6.28, p = 0.041; regression coefficient = 0.043. (b) Foot attachment (residuals)—R2 = 0.43, F1,7 = 7.06, p = 0.033; regression coefficient = 0.028. (c) Crawling (residuals)—R2 = 0.43, F1,7 = 7.08, p = 0.032; regression coefficient = 0.029). (Online version in colour.)

4. Discussion

The main aim of this study was to investigate whether there was evidence for a heritable component in the timing of embryonic developmental events in Radix balthica. There was clear evidence for such a link, with a positive relationship in the timing of a suite of 12 physiological and morphological developmental events between parents and their offspring, and evidence of heritability in the timing of two events, foot attachment and crawling. Both of these events were shown by Smirthwaite et al. [28] to exhibit heterochronies within a clade of freshwater pulmonates, including differences in event timing at the level of families and species. Indeed, these two events were associated with a heterochrony that separated R. balthica from its congener R. auricularia. The parent–offspring similarity reported here is further supported by previous work with R. balthica embryos freshly collected from the field, which revealed increasing inter-individual differences in developmental timing with increasing genetic distance, suggesting the presence of a genetic component to this inter-individual variation [17]. Our findings lend weight to the idea that intraspecific, and even inter-individual differences in developmental event timing could be the raw material from which heterochronies arise.

Given we have shown that, within a single generation, similarities in developmental timing between parents and their offspring are present and heritable, it could be argued that the notion of heterochrony as a macroevolutionary pattern, restricted to interspecific differences, is too narrow. Historically, agreeing on a definition of heterochrony has been controversial [29]; however, ‘altered developmental event timing between ancestors and their descendants’ is now a widely accepted usage of the term [7,8]. Our findings show that altered timing between ancestors and descendants (heterochrony) is not only interspecific, but can also be observed at the intraspecific level. Heritable inter-individual developmental event timing could provide a fundamental link between ontogeny and phylogeny, although what these heritable timing differences mean at the organismal level is still unclear. An important challenge remains in understanding processes occurring in developmental event timing at a micro-evolutionary scale (e.g. heritability, developmental plasticity and epigenetics) within the context of heterochrony as a macro-evolutionary pattern, and this understanding will only be achieved by studying developmental event timing across this continuum (micro- to macro-evolution).

Outcomes of altered timing were categorized by Richardson et al. [9], in a review of vertebrate limb development as: (i) functional changes in the adult; (ii) functional changes in the embryo (without change in the adult); and (iii) changes not related to selection for an adaptive trait. These authors warned against creating adaptive scenarios for heterochrony based on examples of timing shifts that seem to correlate with either changes to development or environment. Identifying whether the altered event timings that we identified translate to difference in function in either the embryo or adult will be an important future step in revealing how selection might act on variation in the timing of developmental events. In the study that documented heterochrony in the group of pond snails to which R. balthica belongs, Smirthwaite et al. [28] discussed that in the two physid species they studied, foot attachment and crawling occurred markedly early, relative to other species. Work using adult physids has revealed that this group has a pronounced ability to use crawling as an escape response from predation [30]. Early crawling in embryonic physids could perhaps therefore lead to improved crawling function in adults in line with scenario (i) described by Richardson et al. [9]. Moreover, embryonic Lymnaea stagnalis, a species from the same family as R. balthica, is known to exhibit increased spinning frequency in response to hypoxia [31], which has been suggested to improve mixing of capsular fluid, providing an adaptation to the embryo for surviving hypoxic environments [32]. Here, we see a parallel with scenario (ii) outlined by Richardson et al. [9].

The variation we observed occurs in a diverse array of developmental events and, therefore, it is conceivable that altered timing of particular events might result in any of the three functional outcomes outlined by Richardson et al. [9]. It is also probable that the consequences of such altered event timing might be very environment-specific. A potentially fruitful future research direction for work investigating intraspecific altered timing will be to consider the implications of how environmentally induced altered timing contributes to its heritability. Research into such developmental plasticity, including epigenetic effects and incorporating next-generation sequencing approaches, could provide significant advances in our understanding of heritable developmental event timing within an ecological context and the links between environment, ontogeny and phylogeny.

Acknowledgements

This study was supported by a studentship to OT from Plymouth University. The development of the bio-imaging technology was supported by Plymouth University's Marine Institute, Research and Innovation Department and Faculty of Science and Technology. We thank Ann Torr for her technical assistance in the laboratory and Andy Foggo for advice on statistical analysis.

  • Received June 7, 2013.
  • Accepted July 29, 2013.

References

View Abstract