Rapid declines in metabolism explain extended coral larval longevity
- First Online:
- Cite this article as:
- Graham, E.M., Baird, A.H., Connolly, S.R. et al. Coral Reefs (2013) 32: 539. doi:10.1007/s00338-012-0999-4
- 491 Views
Lecithotrophic, or non-feeding, marine invertebrate larvae generally have shorter pelagic larval durations (PLDs) than planktotrophic larvae. However, non-feeding larvae of scleractinian corals have PLDs far exceeding those of feeding larvae of other organisms and predictions of PLD based on energy reserves and metabolic rates, raising questions about how such longevity is achieved. Here, we measured temporal changes in metabolic rates and total lipid content of non-feeding larvae of four species of reef corals to determine whether changes in energy utilization through time contribute to extended larval durations. The temporal dynamics of both metabolic rates and lipid content were highly consistent among species. Prior to fertilization, metabolic rates were low (2.73–8.63 nmol O2 larva−1 h−1) before rapidly increasing to a peak during embryogenesis and early development 1–2 days after spawning. Metabolic rates remained high until shortly after larvae first became competent to metamorphose and then declined by up to two orders of magnitude to levels at or below rates seen in unfertilized eggs over the following week. Larvae remained in this state of low metabolic activity for up to 2 months. Consistent with temporal patterns in metabolic rates, depletion of lipids was extremely rapid during early development and then slowed dramatically from 1 week onward. Despite the very low metabolic rates in these species, larvae continued to swim and retained competence for at least 2 months. The capacity of non-feeding coral larvae to enter a state of low metabolism soon after becoming competent to metamorphose significantly extends dispersal potential, thereby accruing advantages typically associated with planktotrophy, notably enhanced population connectivity.
Marine invertebrate larvae can be broadly classified into two categories, lecithotrophs or planktotrophs, depending on their source of nutrition during development. Planktotrophic larvae require external food sources to complete development, whereas lecithotrophic larvae are capable of completing development based solely on maternal provisions (Thorson 1950). The potential to feed should enable planktotrophic larvae to survive longer in the plankton (Scheltema 1986), and there are numerous examples of planktotrophic larvae that spend more time in the plankton than closely related species with lecithotrophic development (e.g., Emlet et al. 1987; Kempf and Todd 1989; Shanks et al. 2003). The longer PLDs of planktotrophic larvae are thought to confer greater dispersal potential for species with such larvae, enabling higher levels of gene flow over larger areas and potentially larger geographic ranges compared to species with non-feeding larvae (Jablonski and Lutz 1983; Pechenik 1999). Relationships between PLD, genetic population structure and range size have been documented for echinoids (Hunt 1993; Emlet 1995), gastropods (Hoskin 1997; Collin 2003; Paulay and Meyer 2006), and various other invertebrates (Foggo et al. 2007; Selkoe and Toonen 2011). However, the role of PLDs in driving such relationships is not always clear (Weersing and Toonen 2009), because population connectivity depends on many factors, including post-settlement processes (Marshall et al. 2010) and barriers to dispersal (Keith et al. 2011), that may obscure the role of larval duration alone.
In reef-building scleractinian corals, larval development mode is generally a good predictor of patterns of dispersal and connectivity. Populations of brooding species, whose larvae are ready to settle on release, typically have higher genetic structure than broadcast-spawning species, whose larvae have an obligate planktonic period of 2–4 days (e.g., Hellberg 1996; Nishikawa et al. 2003). Corals, along with some high-latitude echinoderm taxa, are the only groups with non-feeding larvae for which extremely long PLDs have been documented (Birkeland et al. 1971; Hartnoll 1975; Sebens 1983; Bosch and Pearse 1990; Bryan 2004; Hizi-Degany et al. 2007; Graham et al. 2008; Connolly and Baird 2010). For most lecithotrophs, reported PLDs are short, ranging from hours to days for sponges (Maldonado 2006) and bryozoans (Wendt 2000) to a month for molluscs (Onitsuka et al. 2010) and echinoderms (Villinski et al. 2002). However, coral larvae can survive up to 200 days (Graham et al. 2008) and can complete metamorphosis up to at least 100 days after spawning (Hizi-Degany et al. 2007; Connolly and Baird 2010). These examples indicate that many coral species have developed strategies to extend larval duration and thus accrue the advantages of dispersal traditionally associated with planktotrophy.
If lecithotrophic larvae are to survive long periods in the plankton, they must possess a large supply of stored energy, have low metabolic rates or be able to supplement their endogenous reserves. There is some evidence that echinoderms with lecithotrophic larvae provision their offspring with more energy than their relatives with planktotrophic larvae (Moreno and Hoegh-Guldberg 1999; Falkner et al. 2006; Prowse et al. 2008), but comparative studies for other taxa are lacking (Moran and McAlister 2009). For scleractinian corals, a few species equip their propagules with photosynthetic symbionts (zooxanthellae), which may provide energy to larvae during dispersal and support PLDs over 100 days (Richmond 1987; Harii et al. 2010). However, most (>75 %) coral species have larvae that lack such symbionts (Baird et al. 2009), yet even larvae of these species have exceedingly long PLDs. In Acropora tenuis larvae, initial energy content and metabolic rates observed during the first few weeks after fertilization imply larval longevities of only ~30 days (Richmond 1988; Graham et al. 2008). This estimate is less than half the 69 days observed for this species (Nishikawa et al. 2003), and many months less than larval longevities reported for other Acropora species (Graham et al. 2008; Connolly and Baird 2010). Similar discrepancies between energy reserves, metabolic rates and observed larval durations have been found in echinoderms (Bryan 2004). This suggests that corals and at least some other lecithotrophic larvae must either reduce their metabolic rates substantially as they age, take up additional energy [e.g., by absorption of dissolved organic matter (DOM)], or combine elements of both of these strategies. Although there is some evidence for uptake of DOM in soft coral larvae (Ben-David-Zaslow and Benayahu 2000), this has not been documented in scleractinian coral larvae. Similarly, knowledge of metabolic rates for coral larvae is limited. Temporal changes in lipids over the first month after fertilization suggest that lipids are depleted rapidly in the first week, after which the rate of depletion slows and then picks up again in the fourth week (Harii et al. 2007; Figueiredo et al. 2012). Consistent with these patterns of lipid depletion, respiration rates of Acropora intermedia larvae declined to about one-third of peak values 1 week after spawning (Okubo et al. 2008). These studies indicate that metabolic rates can decrease as coral larvae age, although these changes do not appear to be of sufficient magnitude to account for the extended PLDs documented in coral larvae.
Determining whether long PLDs are widespread among coral species with lecithotrophic larvae, and understanding how these PLDs are attained, is critical to estimating the dispersal potential of this dominant group of reef builders. Such knowledge has important implications for understanding the ecology, evolution, and biogeography of corals. Therefore, we investigated physiological mechanisms underpinning extended PLDs in scleractinian corals with non-zooxanthellate, lecithotrophic larvae, by quantifying respiration rates and energy use over larval life spans. We found strong evidence in both respiration rates and lipid levels for a rapid decline in rates of energy use within approximately 1 week of spawning, even though larvae are still capable of metamorphosis. We conclude that low metabolic rates (i.e., hypometabolism) allow non-feeding coral larvae to extend larval life by minimizing depletion of their energy reserves.
Materials and methods
Study site and larval cultures
The study was conducted at Orpheus Island, Australia, in December 2008, and November and December 2009. Gametes from a total of four broadcast-spawning scleractinian species whose larvae lack zooxanthellae (Goniastrea aspera, A. tenuis, Acropora nasuta, and Acropora spathulata) were collected and cultured using established methods (Willis et al. 1997). Four to six adult colonies of each species were collected immediately prior to anticipated spawning dates and brought onshore. For each species, gametes were collected within an hour of release and combined. Once fertilized, developing embryos were transferred to 500 l fiberglass aquaria with flow through 0.2-μm filtered seawater (FSW) and continuous aeration, one tank per species. We adopted this approach (four species, with one tank per species, rather than, for example, four tanks of a single species) because we were interested in the commonalities in the pattern of energy use in scleractinian corals; thus, we emphasize replication across species rather than within species. However, the absence of within-species replication across spawning events or tanks means that species differences cannot be separated from cohort-specific or tank effects. Therefore, we do not test for differences between species. The aquaria were maintained in temperature controlled rooms at near-ambient temperature (27 ± 1 °C) and a 12-h light/dark cycle.
At regular sampling intervals, subsamples of eggs, embryos or larvae (hereafter “propagules”) were randomly selected, for each species, and used for respiration measurements and lipid analysis. The first sample was taken from newly released gametes, prior to fertilization. For respiration measurements, five replicates of 50 propagules were used. For lipid analysis of each Acropora species, the same 50 propagules used to measure respiration rates were subsequently frozen and used for lipid analysis. However, for lipid analysis of G. aspera propagules, due to their smaller size, three replicates of 600 propagules were used. Sampling took place every 12 h for the first 36–48 h to capture larval development through embryogenesis to a swimming larva, followed by daily sampling until the majority of larvae were competent to settle at 5 days after spawning. From 5 days after spawning, sampling was further reduced to every 3 days, unless low numbers of surviving larvae forced a further reduction to weekly sampling. Survival experiments and settlement assays, described in detail below, were conducted at each sampling point after larvae began swimming to determine whether patterns in energy use affected larval mortality rates or the larvae’s capability to metamorphose.
To measure egg, embryo, and larval respiration rates, we used a temperature compensated, fiber-optic oxygen meter (Fibox, PreSens GmbH). Respiration chambers were custom made, with each chamber consisting of a 1.5-ml glass vial integrated with a 5 mm diameter oxygen sensitive sensor foil spot on the inside of the chamber. Prior to each experiment, the respirometer was calibrated using sodium dithionite (Na2S2O4) and air-saturated FSW for a two-point (0, 100 %) calibration following the manufacturer’s instructions. A sterilized miniature magnet was placed inside each chamber and magnetically stirred to ensure adequate mixing. For each of the five replicates of each species, 50 propagules were counted into the chamber and topped with fresh 0.2 μm FSW. The change in oxygen concentration in the chamber was measured for 5 min. Following each replicate, the chamber was flushed and refilled with fresh FSW, and oxygen measurements were taken of the individual-free seawater for an additional 5 min to serve as a control. Oxygen consumption was calculated as the slope of oxygen concentration over the 5 min measuring period and then converted into nmol O2 larva−1 h−1.
There is no a priori theory that predicts a particular functional form to describe how oxygen consumption should change with days after spawning. Moreover, initial plots of oxygen consumption rate as a function of days after spawning suggested a complex nonlinear relationship between the two variables. A log transformation of the observed values improved the homogeneity of variances, but the underlying relationship remained highly nonlinear. Therefore, to analyze the change in oxygen consumption rate, a nonparametric generalized additive model (GAM) was fitted to the log-transformed data. GAM uses a locally weighted smoothing function to characterize arbitrary nonlinear relationships between the response (respiration rate) and predictor (days after spawning) variables (Zuur et al. 2009). GAM was implemented using the mgcv package in R (Wood 1994).
To measure the total amount of lipid in each sample, a TLC–FID detection system was used (Iatroscan MK-5). Lipids were extracted using a modified Bligh-Dyer chloroform:methanol method, with an internal standard added to provide an estimate of lipid recovery (Sewell 2005). Two developments were used to separate the lipid sample into the following classes: aliphatic hydrocarbons, wax esters (WE), triacylglycerides (TG), free fatty acids, free aliphatic alcohols, cholesterols (ST), and phospholipids (PL) (Parrish 1999). The lipid classes were then divided into two groups—energetic lipids (WE and TG) and structural lipids (ST and PL)—and analyzed separately. Since energetic lipids are most likely to be available for maintenance of metabolism during the larval phase, we present the energetic lipids in the results. However, we also assess the extent to which qualitative trends in total lipids reflect those of energetic lipids.
Like the respiration data, there is no a priori reason for favoring a particular mathematical function to describe lipid depletion over time. Moreover, visual inspection of lipid data revealed a complex nonlinear relationship with time. Therefore, after applying a square-root transformation to the lipid data (to homogenize variances), a GAM was used to characterize the nonlinearity in the change in lipid levels over time. Our approach was similar to that described above for respiration. However, because coral larvae are non-feeding and lack zooxanthellae, a monotonicity constraint was applied to the GAM (i.e., we constrained fitted lipid levels to decrease over time, as per Wood 1994). This helped to avoid over-fitting of the model and yielded narrower confidence intervals than an unconstrained fit. We obtained 95 % confidence intervals for this constrained GAM fit by bootstrapping residuals (Efron and Tibshirani 1993).
To determine whether larval survival was affected by energy use, once swimming larvae had developed, five replicate 70-ml specimen jars were set up containing 0.2 μm FSW and 100 larvae each. At each sampling time, the surviving larvae were censused and transferred into new specimen jars. Because coral larvae typically lyse within 24 h of death, there was no need to distinguish between live and dead larvae, that is, the larvae remaining at each interval were assumed alive (Baird et al. 2006). A Kaplan–Meier product-limit analysis was used to obtain nonparametric estimates of the median survival time and 95 % confidence intervals around this estimate for each species.
To determine the onset of competence and whether this capability was maintained throughout larval duration, at each sampling point following the onset of swimming, a subsample of 120 larvae was placed into a 6-well plate. Twenty larvae were introduced into each well containing 0.2 μm FSW and a small piece of crustose coralline algae (CCA), a known settlement inducer for Acropora species (Morse et al. 1996). We selected a particular CCA morph from among the locally abundant species around the study site that we know from past work induces settlement. After 24 h, the number of larvae that had successfully metamorphosed was recorded. Metamorphosis was defined as the deposition of a basal plate following Baird and Babcock (2000).
Temporal changes in lipid content and respiration rates of coral larvae over the first month of the larval duration indicate that rates of lipid depletion and respiration can decrease as coral larvae age (Harii et al. 2007; Okubo et al. 2008; Figueiredo et al. 2012), although previously reported changes appear not to be of sufficient magnitude to account for the extended PLDs documented in coral larvae. Here, we document much greater declines in metabolic rates of lecithotrophic, non-zooxanthellate larvae than have been recorded in previous studies. Moreover, patterns of striking declines in the rates of both larval respiration and lipid depletion were qualitatively similar among all four broadcast-spawning species studied in the first 3 weeks after spawning. Although the specific values of oxygen consumption varied among species, in all cases, respiration rates were relatively low for eggs, followed by a rapid increase through embryogenesis to a peak, the timing of which varied among species from 12 to 48 h after fertilization. This spike in respiration rates is short-lived, and rates quickly fall back to initial levels within 3–6 days, where they remain for up to 8 weeks. Consistent with these observations, lipids are depleted rapidly during development, until a few days after larvae become competent to metamorphose, at which time the rate of lipid utilization slows dramatically. The high concordance between these two measures of energy use strongly supports the hypothesis that an extended period of reduced larval metabolism explains, at least partly, the long PLDs observed in many coral species. Moreover, the capacity of larvae to maintain competence throughout this extended period of reduced energy use implies the capacity to settle over a broad range of dispersal distances.
Although embryogenesis and larval development are the most energetically demanding periods in the larval life of many marine invertebrates (e.g., Shilling and Manahan 1994; Anger 1996; Hoegh-Guldberg and Emlet 1997; Bryan 2004), the magnitude of the declines observed in some species suggests that lecithotrophic larvae have the capacity to achieve much lower levels of energy use than have previously been documented. High rates of respiration correspond firstly with high rates of cell division during embryogenesis (approximately 12–36 h post-fertilization) and secondly with the development of specialized cells, such as spirocysts, that are associated with attachment and metamorphosis (36–96 h post-fertilization) (Hayashibara et al. 2000; Okubo and Motokawa 2007). Once these energetically demanding processes are complete, larvae enter a state of substantially reduced metabolism. The magnitude of decrease in respiration varied from ~2.5-fold in A. spathulata to about 100-fold in A. tenuis. Temporal changes in respiration have been examined previously in only one species, A. intermedia (Okubo et al. 2008), and were found to decrease by ~threefold, which is within, but toward the low end of the range of declines in the species studied here.
The pattern of rapid decline in lipid content during the first week of embryogenesis and larval development, followed by a period when further lipid depletion was minimal, is consistent with the observed trends in respiration rate. Overall, larvae lost between half (A. spathulata) to ~75 % (A. tenuis) of their initial lipids during the first week. This is similar to, but larger than, the ~30–40 % depletion of lipids observed in previous studies that report comparable data (Harii et al. 2007; Figueiredo et al. 2012). For at least three of the four study species, lipid levels were stable after this period. The exception was G. aspera, which exhibited a secondary decline over the final (fifth) week. However, this decline must be treated with some caution, because it was driven entirely by samples on the final sampling date, and lipid levels were very stable over the preceding 3 weeks. For the three Acropora species, the rates of lipid depletion apparent at the end of the study are consistent with very long PLDs observed in other Acropora coral species, such as A. valida, which has been observed to successfully metamorphose at ~110 days (Connolly and Baird 2010).
Estimates of lipid remaining after an additional 100 days in the plankton for three Acropora species, assuming maintenance of the lipid utilization rate observed at the end of the study (i.e., change over the last two sampling points)
Second to last sampling point
End of study
Lipid remaining (μg)
Second to last sampling point
End of study
Estimated rate of lipid used (days−1)
Next 100 days
Lipid used (μg)
Percent of remaining lipid used (%)
Estimates of pelagic larval duration (PLD) for four study species using lipid content and oxygen consumption converted to their energetic equivalents [39.5 kJ g−1 lipid; 441 kJ mol O2−1; (Gnaiger 1983)]
O2 consumption (nmol/n/h)
O2 consumption (mJ/n/h)
Lipid remaining (μg)
Energy remaining (mJ)
Estimated remaining (days)
Age + Est. remaining (days)
Maximum observed PLD (days)
Consistent with previous work, wax esters were found to be the dominant lipid class (70–90 %) comprising coral eggs in this study. Coral eggs are provisioned with an unusually large amount of wax esters, particularly when compared to eggs of other non-feeding marine invertebrates, which contain triacylglycerides as the dominant energetic lipid (Moran and Manahan 2003; Prowse et al. 2008). In fact, while triacylglycerides do occur in coral propagules that also contain zooxanthellae, they are only present in very small amounts in coral eggs and larvae without zooxanthellae (1–2 %, Arai et al. 1999, 0–0.4 % Harii et al. 2007; 0.3 % Figueiredo et al. 2012). In this study, the amount of triacylglycerides was also negligible (0.01–0.04 %), and thus both classes of energetic lipids were combined and analyzed together (see “Materials and methods” section). The large quantity of wax esters in coral eggs may indicate a particular provisioning strategy, possibly for buoyancy to enhance fertilization (Arai et al. 1993), or they may be a by-product of adult physiology (20–50 % of total lipid in adult corals is wax esters, Harland et al. 1993; Yamashiro et al. 1999; Imbs et al. 2010). In addition to providing energy to the developing larvae, they may also support the longer planktonic durations of coral larvae or even be used for metamorphosis or post-settlement needs.
The consistency in the overall patterns of reduced oxygen consumption and lipid depletion for all four study species is striking and implies that the energetic cost of delaying metamorphosis may be much smaller than is commonly assumed for lecithotrophic larvae. These findings help to explain large discrepancies between energetic estimates of larval duration based on metabolic rates measured early in larval life (Richmond 1988) and the much greater durations measured empirically (Graham et al. 2008; Connolly and Baird 2010). This suggests that very low basal metabolic rates underpin the extended competence periods and larval durations of scleractinian corals, which are on par with or greater than those of most planktotrophs (Shanks et al. 2003; Shanks 2009). While there are undoubtedly costs associated with increased time in the plankton, our results suggest that some of the hypothesized post-settlement costs, such as increased mortality and decreased growth (Pechenik 2006), may be less severe for scleractinian coral larvae than might be expected based on the metabolic rates prevailing early in larval life or typical of similar-sized invertebrates.
The lipid content of subtropical and temperate invertebrates, including molluscs and echinoderms, generally decreases during larval duration, even for planktotrophic species. Although the decline is typically relatively linear (e.g., Moran and Manahan 2003; Bryan 2004; Sewell 2005; Byrne et al. 2008), Kattner et al. (2003) also found a much greater decrease in lipid during early development for two lecithotrophic crabs compared to later stages, similar to that documented here. Respiration rates, on the other hand, tend to either increase throughout larval duration (e.g., Dobberteen and Pechenik 1987; Hoegh-Guldberg and Manahan 1995; Hoegh-Guldberg and Emlet 1997; Marsh et al. 2001; Moran and Manahan 2004) or rapidly reach a plateau where they remain until the end of the study (e.g., Hoegh-Guldberg 1994; Moreno and Hoegh-Guldberg 1999; Moran and Manahan 2003). Although a decrease in respiration rates after competence was acquired occurred in the lecithotrophic echinoid Heliocidaris erythrogramma (Hoegh-Guldberg and Emlet 1997), whether metabolic rates would remain low is unknown as the study ended after 6 days. In fact, few studies of the temporal dynamics of metabolism in other lecithotrophic larvae have lasted more than 2 weeks (e.g., Okubo et al. 2008; Hoegh-Guldberg and Emlet 1997; Moran and Manahan 2003). This raises the possibility that other taxa with lecithotrophic larvae may adopt strategies similar to those in corals, in order to extend larval longevity and increase energy reserves available for metamorphosis and growth after long-distance dispersal.
E.M.G. thanks Shane Blowes, Karen Chong-Seng and staff from Orpheus Island Research Station for field assistance; Angela Little and Erica Zarate for assistance with lipid analysis; and Loic Thibaut for help implementing monotonically constrained GAMs. This study was partly funded by an award to E.M.G. from the ARC Environmental Futures Network Early Career Researcher Support Program Round 8 Funding. Additional funding for this project was provided by the Australian Research Council through the ARC Centre of Excellence for Coral Reef Studies.