We performed an experimental investigation of larval settlement behaviour for a broad taxonomic cross-section of Indo-Pacific coral species and found that precompetency periods ranged from about two to six days and duration increased with egg size. We also confirmed that extended competency windows (+ 70 d) are possible for at least some species, and identified novel and complex temporal dynamics in settlement behaviour during the competency window that may facilitate long-distance dispersal success. These patterns challenge the long-held assumption that competency gradually wanes over time, with likely significant implications for population connectivity and metapopulation dynamics. Finally, we identified a broadly effective settlement cue in reef rubble, providing a starting point for further investigations into cultivating potential universal inducers.
Precompetency period
Implications of variation in precompetency for dispersal
Precompetency durations ranged 3-fold across taxa, with the Merulinid Goniastrea retiformis on one end of the spectrum at 2.1 DAF, and the staghorn Acropora austera on the other end of the spectrum at 6.1 DAF. Figueiredo et al. 8 modelled time to competency for six broadcasting coral species and found that estimates ranged from 1.4 to 3.8 days. Of the six species Figueiredo et al. 8 tested, three were again tested in this study and our estimates were remarkably similar: we estimated precompetency periods of 2.1, 4.3 and 3.1 d, compared to 1.4, 3.5 and 2.9 d estimated for G. retiformis, A. millepora and P. daedalea, respectively. Using these precompetency periods, Figueiredo et al. 8 modelled the proportion of larvae retained, based on water residence times on the GBR, and found that an increase in precompetency of as little as 1 day led to significantly reduced local larval retention. Therefore, our results, which quantified precompetencies ranging from 2.1 to 6.1, suggest that larval retention should vary widely amongst coral taxa. For example, if local-scale currents are moving at 10 cm s− 1, a 4-day increase in precompetency would result in an additional 35 km of transport.
Egg size has previously been correlated with time to motility 8 and propagule size has been positively correlated with precompetency duration in brooding corals 64. Our results support this relationship and expand this to more taxa and reproductive strategies, indicating that larger embryos take longer to reach settlement competency. The significant relationship between oocyte diameter and TC50, therefore, may be used to estimate TC50 for other taxa for which competency data are unknown (Fig. 3A).
Taxa-specific patterns and their implications for community composition
High variation in precompetency allowed us to classify species as having either short (≤ 3 d), mid (> 3–5 d), or long (> 5 d) precompetency periods (Fig. 3B; Table 1) and revealed notable taxa-specific patterns: most Merulinids, including the genera Goniastrea, Dipsastraea (formerly Favia), Platygyra, and Oulophyllia had ‘short’ precompetency periods while all ‘long’ precompetency-period species were of the genus Acropora. ‘Mid’ precompetency durations were most common and included species from the greatest diversity of families (5). While these differences could potentially result in comparatively greater local retention in Merulinids and greater dispersal of Acroporids, it is difficult to determine whether this is reflected in population genetic structure for several reasons. Firstly, small spatial-scale genetic structure is rarely studied in corals 65, and these studies primarily investigate brooding species with extremely short pelagic larval durations (< 1 d, i.e. 66,67), or compare brooding and broadcasting reproductive strategies (i.e. 64). Therefore, little data exist directly comparing genetic diversity and dispersal capacity across space and taxa of broadcast spawners. Secondly, genetic structure within a local population can vary between recruits and adults, indicating that the adult populations sampled don’t necessarily reflect larval connectivity 67. Furthermore, broadcast spawning taxa with similar abilities to disperse can display different genetic structure (i.e. 68), with post-recruitment mortality processes also contributing to population structure. Despite these challenges, data to date would suggest that most recruitment is through local retention, but that enough propagules disperse long distances to ensure metapopulation connectivity across distant reefs 8,64,69,70. Indeed, the modelled TC50 values were not significantly correlated with geographical species distributions, as estimated by Hughes et al. 71, (Figure S 2), suggesting that this variation in precompetency doesn’t necessarily correlate with biogeographic distributions on evolutionary time scales; the lack of a correlation is likely due to a number of additional factors influencing connectivity including the degree of isolation, local population size, oceanography, and the competency window.
Importance of within-cohort variation in precompetency
High within-cohort variation in precompetency is characteristic of spawning corals 2 and other invertebrates 63 and is influenced by many factors including the genetic diversity within the cohort 63 and the environmental conditions during larval rearing, which affect rates of larval development 9,72. Our analysis applied a cohort-level threshold of 0.3 as the definition of ‘competent’ and thus the estimates here are most useful for understanding the onset of competency and, consequently, self-seeding dynamics 2. On the other hand, the TC50 estimates modelled with a threshold of around 0.7 (Fig. 2B, Fig. 3B) would be most useful for understanding long-distance dispersal potential over evolutionary time scales 64. While there was some variation in TC50 estimates amongst the thresholds used, the rank order of taxa was fairlyconsistent (Fig. 3B) suggesting that the gradual attainment of competency within the cohort occurs similarly amongst taxa.
Temporal patterns in competency
Our study precluded a comprehensive investigation of the timing of loss of competency. In many species tested, however, larvae remained competent until the final available timepoint and some taxa demonstrated the potential for extended competency during the PLD (Fig. 4). These results corroborate the findings of Graham et al. 3,73, Harrison et al. 74, Wilson and Harrison 75, and others who describe competency windows in excess of 100 days for some taxa (Table S 1), and suggest that larval connectivity may extend to greater distances than initially estimated for some reef systems, such as those in Western Australia 76.
Interestingly, larval settlement fluctuated through time with peaks and troughs that were often consistent amongst cues within a species, and across taxa. This consistency suggests that fluctuations in settlement are likely reflective of larval behaviour and physiology and not cue variability alone, thereby challenging the assumption that competency gradually wanes over time. Periods of inactivity during the competency window, such as was seen in the bimodal pattern of every species tested beyond 40 days (A. hyacinthus, D. matthaii, D. pallida and Mycedium elephantotus) (Fig. 4) could represent a ‘bet-hedging strategy’ 77, where a pulse of local settlement is followed by a metabolically inactive period of pelagic dispersal facilitating connectivity to non-natal areas for colonization when settlement competence increases once more. This is supported by evidence from Graham et al. 73 indicating that larvae can enter a state of low metabolic activity shortly after becoming competent to settle, supporting the capacity for long-distance dispersal. The ‘desperate larval hypothesis’—the notion that larvae become less discriminatory as they age 30—may also explain the resurgence in settlement behaviour at later timepoints; indeed, some species settled in response to more cues during later timepoints (i.e. D. matthaii and D. pallida), while the prevalence of indiscriminate settlement—settlement in the absence of any cue—also increased through time (Fig. 4). Yet whether this temporal variation has realized consequences for dispersal also depends heavily on survival throughout the pelagic period (Fig. 5).
Larvae weren’t fed in this study but there is mounting evidence that at least some species of coral larvae are capable of heterotrophic feeding 78 and the uptake of dissolved organic matter to supply amino acids 79. Many larvae can also ingest symbionts through their mouth and incorporate them into their endoderm 80,81, and eggs of vertical transmitters host symbionts from parental colonies. Indeed, all vertical transmitters tested in this study (P. cylindrica, P. lobata, M. digitata, and M. aequituberculata) showed active settlement at the final time point (22–32 days) suggesting that nutrition from symbionts may aid in supporting long competency windows. Yet many horizontal transmitters also demonstrated extended competency in the absence of feeding. Further research is needed to investigate whether heterotrophy and symbiont uptake during the larval stage can influence these durations, and therefore, metapopulation connectivity.
Inter- and intra-specific patterns in response to settlement cues
Reef rubble was the most effective settlement cue for 76% of species tested and was significantly better at inducing settlement than the CCA and biofilm disc in nearly all non-Acropora species. Less settlement on CCA and live biofilm discs compared with rubble fragments has several possible explanations that are not mutually exclusive. Firstly, the physical features of complex microhabitats within rubble may be an important consideration in a larva’s settlement decision. Larvae are known to preferentially settle in microhabitats that offer refuge from external pressures 46,55,82–84 and to seek out habitats with lower light conditions 85. Rubble fragments were texturally complex and offered ample microrefugia of various sizes that likely attracted larvae. Secondly, rubble fragments may have supported more complex and/or well-developed biofilms than the artificial discs, leading to the presence of stronger induction cues. Indeed, the presence of complex microhabitats likely created pockets of unique biofilm communities (Turnlund et al. in review) that may have offered a wider diversity of potential inducers. It is also likely that biofilm discs supported early successional species (+ 4 wk conditioned) 52 that were less mature and inductive than those on rubble fragments. Finally, chemical extracts of some dead coral and CCA skeletons can induce coral larval settlement, indicating the potential presence of legacy inducers within these calcareous matrices 51 and this may also be the case for the reef rubble applied in the current experiment. Regardless of the mechanism, reef rubble was overwhelmingly the best non-Acropora settlement cue, highlighting the importance of chemical characterisation of rubble substrates to identify potential inducers
The best settlement inducer for Acroporidae varied by species and rubble was also a strong inducer across Acropora. This result was somewhat surprising because CCA are known to be strong settlement inducers in Acropora25–28,56. However, Abdul Wahab et al. 27 recently demonstrated species-specific preferences amongst coral/CCA pairings, and Porolithon was not the most universal CCA cue. Therefore, other CCAs may be as effective as rubble at inducting settlement across Acropora spp. but this remains untested.
The GLW-amide peptide Hym-248 50,86 significantly induced settlement in all Acroporidae species, but failed to induce settlement in nearly all non-Acropora species (Fig. 4; Figure S 1), corroborating the findings of Erwin and Szmant 51 and indicating that this signalling pathway is likely not conserved amongst taxa within the Scleractinia, with high specificity of neuropeptide activators at low taxonomic levels. The testing of additional neuropeptides across a wider range of concentrations will improve our understanding of larval neurobiology.
Limitations
While this laboratory-based experimental design was effective for assessing larval development and function, there are limitations to assessing larval behaviour in small-scale experiments. For example, larvae were confined in close proximity to physical and chemical cues for settlement and not offered a choice. Thus, comparisons of absolute settlement success between cues should be interpreted cautiously. By contrast, the strategy of offering individual cues to identify the shortest precompetency period is an effective way of ensuring competency periods for larvae of diverse taxa are compared fairly. Indiscriminate larvae are more likely to settle soon after reaching competency, while selective species may spend longer in the plankton before encountering their preferred cue 63. Because the preferred cue of most coral species is unknown, it is possible that we overestimated precompetency for those species that are selective (i.e., optimal inducers for settlement may not have been offered). Our data also indicated that at least some species reach peak competency at or near 100%, while others don’t reach complete competency in response to any of the provided cues. The reasons for not reaching 100% competency could be related to many factors including species- or genotype-specific differences in maximum competence, sub-optimal larval health, or that none of the cues selected were optimal for that species 27. More work is needed to understand this phenomenon. Furthermore, we tested single larval cohorts for most taxa (Table S 2). Yet high within- and between-cohort variation in competency windows 2 has the potential to shift TC50 estimates. Thus, when applying these results to models of local retention and dispersal, it would be prudent to embed uncertainty around the estimates (Fig. 2A).
Lastly, it wasn’t possible to completely standardize the physical and biological settlement cues due to inherent variability in natural substrates and variations in spawning and settlement times. We attempted to account for this by: (i) using consistent (i.e., the same ‘parent’) rubble and CCA fragments across all treatments and species within a given month whenever possible; (ii) conditioning all substrates together, (iii) haphazardly loading substrates across all experimental wells to minimize bias, (iv) creating standard-sized substrates, and (v) applying robust replication. Despite this, there was variability in larval responses within treatments; whether this variability was related to changes in the quality or potency of the cue, or behavioural changes throughout ontogeny is difficult to untangle. Yet, overall patterns in settlement behaviour were remarkably consistent through time amongst treatments, and across species, supporting the patterns we describe.
Significance in a changing climate – Applications for reef restoration
Human activity and global climate change is fundamentally altering connectivity and recovery processes 87. For example, declining adult densities can reduce larval supply directly 38, as well as indirectly via allee effects 43. In addition, warmer temperatures can accelerate larval development 9,72,88, likely reducing the precompetency period and the maximum pelagic larval duration 89, leading to increased retention and reduced downstream dispersal and connectivity. The accompanied changes and shifts in ocean currents can also influence local retention and downstream dispersal of corals. At the reef scale, shifts in the benthic-community composition driven by marine heatwaves and acidifying conditions can impact the quality and quantity of larvae, their settlement cues, and the inducer-inhibitor ratio of organisms on the seafloor 90,91. Yet the specificity for larval settlement cues 26,56,92, the duration of precompetency 10, and the PLD 3,11,93,94 vary considerably amongst species. Therefore, climate change is likely to unevenly affect the dispersal and recovery potential amongst species, resulting in broad-scale and species-specific implications for biogeographic patterns in coral-community composition, the colonisation of new habitats, and the expansion of species ranges under climate change.
There are four important management and restoration impacts resulting from this multi-species coral settlement study. The first is an appreciation of the variability within cohorts and between coral species. This variability may serve as a natural insurance policy against disturbances where local-retention and downstream connectivity are likely co-occurring and contributing to reef resilience. Second, the efficiency of seeding sexually produced corals for use in reef restoration purposes 43 can improved through refined timing and prioritisation of settlement by taxa based on our data, saving time and costs while maximising output from aquaculture facilities. Third, the new knowledge of extended competency durations allows for the staged settlement and grow-out of spat prior to deployment, adding efficiencies and maximising outputs of the coral production and restoration process. Finally, the multi-species data on precompetency and competency dynamics, and the potential impact on local retention and downstream connectivity will improve the spatial planning process 95 to help manage climate impacts and maximize the success of restoration efforts.