Egg size and fecundity of biannually spawning corals at Scott Reef

Egg size and fecundity are often used as proxies for coral reproductive success and health. The amount of energy a coral invests in reproduction reflects its environmental conditions during gametogenesis. Additionally, assuming resources for reproduction are limited, it is thought that an increase in egg size should result in a decrease in the number of eggs produced i.e. investing in many small eggs or fewer larger eggs. The biannually spawning populations of Scott Reef offer a unique opportunity to compare the egg size and polyp fecundity of corals exposed to different environmental conditions during gametogenesis, prior to spawning in autumn (March) and spring (October). In this study, we investigated the relationship between egg size and polyp fecundity within and between seven Acropora species from 2008 to 2010. We also quantified the fecundity and egg size of four Acropora species that spawn during both autumn and spring (2008–2010). We found no seasonal variability in egg size and fecundity in the species studied here, possibly as a result of a summer light regime being impacted by high cloud cover in cyclone season. There was high natural variability and no apparent trade-off between egg size and fecundity, both within and between each species. These findings challenge the assumption that egg size and fecundity are negatively correlated, or that a simple, energetically constrained trade-off exists between the two.

www.nature.com/scientificreports/ the 1998 bleaching on the GBR, some coral species had no eggs, while others had eggs that were reduced in both size and number 14 . In addition to reducing energy available for reproduction, environmental variation can also cause the available energy to be re-directed to other life history traits. For example, in low light conditions energy was channeled to calcification but not reproduction 16 .
Increasing the size and number of eggs requires the use of energy that has been channeled to oogenesis (egg production). Larger eggs are fertilized at a greater rate, probably because they provide a larger target for sperm 17 . Additionally, larger eggs have more stored lipids 15 , which provide important energy reserves for the azooxanthellate larvae produced by Acropora corals. Until azooxanthellate larvae and new recruits are able to acquire zooxanthellae, they rely on the lipid stores provided by the parent to meet their energetic needs 18 . Consequently, these energy stores affect both larval duration (which affects geographic distribution) and postsettlement survival 18,19 . Conversely, producing more eggs has obvious life history advantages, because it increases the numerical odds of self-proliferation 15 . This suggests that there is a trade-off between producing fewer large eggs, each with an increased chance of survival and greater potential for dispersal, or many small eggs with reduced survival and dispersal 19,20 . An inverse relationship between egg size and fecundity has been reported in several studies [21][22][23] , however, these studies only compare between species and not within a species. The tradeoff or inverse relationship between fecundity and egg size can also change in response to stress. For example, Acropora millepora colonies that were exposed to temperature stress, produced fewer eggs but maintained egg sizes 15 . Similarly, while both fecundity and egg size were reduced in bleached soft corals, energy allocated to the number of eggs appeared to be re-directed to maintaining fewer eggs to reach mature sizes 24 . In this study, we investigated whether there were seasonal differences in egg size and polyp fecundity in biannually spawning Acropora corals from Scott Reef, and whether there was an inverse or trade-off relationship between egg size and fecundity both within and between species.

Results
environmental conditions. Scott Reef is a relatively low latitude reef (14°S), but has some seasonal variation in environmental conditions. In the months leading up to autumn spawning (November to March), the days are longer (12-13 h) and water temperatures are warmer (29-30 °C); in the months leading up to spring spawning (June to October), the days are shorter (11-12 h) and water temperatures are cooler (26-28 °C) (Fig. 1a,h). Cyclone season is from December to April, and the majority of the rainfall occurs during the summer months of January and February, with correspondingly high cloud cover during these months (Fig. 1e,f). Consequently, the overall daily solar exposure is highest in October and November rather that the months with most daylight hours (December and January), and there is relatively little annual variation in overall solar exposure (ranges from ~ 17 to 27 MJ/m 2 ) (Fig. 1d). egg size versus fecundity. Egg sizes ranged from ~ 450 to 800 μm, with means of ~ 600 to 700 μm across all species and both seasons. Egg counts per polyp had high variability within each species, ranging from 4 to 12 eggs but with means ~ 6 to 8 eggs per polyp (Fig. 2). There was a poor correlation between polyp fecundity and mean egg sizes both within and between the seven Acropora species. In all but one species, egg size did not decrease significantly with increased fecundity (Fig. 2). In A. spicifera, mean egg sizes decreased significantly with increased fecundity (P = 0.002), but fecundity explained only ~ 8% of the variation in egg size. Within the other species, the relationship between fecundity and egg size was weak (R 2 = < 0.16) and insignificant (P > 0.05) (Fig. 2). Similarly, between the seven species there was no correlation between mean egg size and mean number of eggs (R 2 = 0.10, P = 0.488) (Fig. 3).
Seasonal effects on egg size and fecundity. Four biannually spawning Acropora species (A. gemmifera, A. hyacinthus, A. microclados and A. tenuis) were used to examine seasonal variability in reproductive output. No differences in mean maximal egg size were observed between the autumn and spring spawning events, across all of the species studied (Fig. 4, Supplementary Table S3). Similarly, there were no differences in the number of eggs per polyp in all species; the exception was A. microclados, which had higher fecundity in spring compared to autumn.

Discussion
Many Acropora corals at Scott Reef spawn biannually, but most individuals spawn either in autumn or in spring (not in both seasons) and are thus temporally isolated from one another with respect to reproduction 10,25 . Gametogenesis takes ~ 4 to 6 months in Acropora corals at Scott Reef 10 . While it is unknown whether gametogenesis occurs at different rates in the different seasons, coral colonies experience different environmental conditions through the gametogenic period, leading up to spawning in the spring and autumn spawning seasons. Gametogenesis occurs through austral winter and early spring prior to the spring spawning (October/November), when water temperatures are cooler and days are shorter. Conversely, gametogenesis occurs through summer prior to the autumn spawning (March/April), when water temperatures are warmest (potentially stressfully warm), the days are longest, and tropical cyclones occur. Both spring and autumn spawning correspond with seasonal minimums in wind speed 26,27 . Within the thermal tolerance limits of the coral, warmer water temperatures and longer days theoretically increase energy availability for reproductive processes through increased metabolic activity and elevated photosynthesis 28,29 . Despite the different environmental conditions through gametogenesis, there were no seasonal differences in fecundity and eggs size observed in the biannually spawning Acropora corals studied here. There are several possibilities for the lack of seasonality observed in reproductive output. Firstly, fecundity and egg size varied widely within species, which confounded inferences about whether reproductive output was higher in a particular season. Other studies have similarly reported high variability, particularly in www.nature.com/scientificreports/ fecundity. Fecundity can vary widely with season 22,30 and between years 31 , but there is also high variation between colonies, within a single colony 22 , with colony age 30 and between colonies at different depths 13 . Fecundity can also vary in response to stressors 13 , however, there was no evidence of environmental stress, such as damaging waves from cyclones or heat stress causing coral bleaching 32 , before or during the period when samples were collected for this study. Adaptive plasticity in egg size (in response to conditions parents are exposed to), is discussed further below. Secondly, Scott Reef is situated in the tropics (14°S) with relatively small seasonal variations in temperature and day length, and has a light regime that is affected by high cloud cover during the summer  www.nature.com/scientificreports/ cyclone season. Water temperatures are 2-4 °C cooler in the winter months leading up to spring spawning than in the summer months prior to autumn spawning. Day length is 1-2 h shorter in winter compared to summer, however summer cyclones and rainfall mean that overall sunshine hours are higher in the winter months. Consequently, there are cooler temperatures with more sunshine hours in the months leading to the spring spawning and warmer temperatures with less sunshine hours in the months leading to the autumn spawning, which may result in comparable available energy for reproduction during both spawning events. Thirdly, seasonal differences in environmental conditions may indeed drive some seasonal differences in energetics, but these could be channeled into other life history processes, such as calcification 12,33 , rather than fecundity and egg size. Variation in available energy may also affect egg quality rather than size or number. For example, in other invertebrates (greenlip abalone), while the size of the eggs do not increase, the density of protein and lipids increase throughout the spawning season 34 and may indicate an increase in the quality over size of the egg. However, a study on the reef-building coral Montipora capitata, reported stable egg quality (lipids and antioxidants) regardless of the environmental conditions the parent colonies were exposed to, although egg sizes were not presented in this work 35 . The higher polyp fecundity in spring observed in A. microclados may have been an adaptive response to cooler (less favourable) conditions in spring. That is, an increase in parental investment to increase survival in less favourable conditions 36 . Alternatively, more sunshine hours during winter gametogenesis may have provided additional energy to produce more eggs in this species. Early work on egg size and number of eggs suggests a simple trade-off model. That is, assuming resources for reproduction are limited, then an increase in gamete size should result in a reduction in the number of gametes 37 . Correspondingly, earlier studies of different coral species, genera and morphologies reported an inverse relationship between coral egg sizes and the number of eggs (fecundity) [21][22][23] , also suggesting that energy is channelled to either fewer large eggs or many small eggs 19,20 . However, in these cases, the reductions in fecundity with egg size among genera were attributed to the differences in polyp morphology (and sometimes reproductive mode i.e. brooder vs spawner). That is, differences in polyp size and structure can also affect egg size and fecundity 38,39 independently of energetics. In our between species comparison, we did not see an inverse relationship between egg size and number of eggs (Fig. 3), but there were also no large differences in corallite size for the seven Acropora species studied here (see Supplementary Table S4 for corallite sizes of our species). However, it is important to note that the differences in reproductive mode and morphology (including polyp structure and size) between genera and species, interferes with the egg size versus number of eggs comparison in the context of a trade-off model. In order to determine if there is a trade-off between egg size and number of eggs, we need to look at individuals within a species. That is, do individuals with large eggs have fewer eggs than individuals with smaller eggs of the same species? We have been unable to locate any other dataset providing a within species comparison. Our study demonstrates that there is no direct relationship between egg size and fecundity, within these species of Acropora, and suggests that there is more than just a simple trade-off in resources influencing these measures.
Egg size has been shown to be a phenotypically plastic trait, regulated by the conditions the parent colony is exposed to. For example, a study on the broadcast spawning ascidian, Styela plicata, demonstrated that parents maintained at high densities produced smaller eggs, presumably reflecting the higher sperm concentrations expected at high adult densities, and therefore reduced requirement for a large target 40 . While the study was unable to measure the number of gametes, and provide an egg size versus number of eggs comparison, it did suggest that egg size is an adaptive plastic response, rather than a simple energetic constraint. Several studies have also reported varying effects of stress on the number and size of eggs. Under temperature stress sufficient to cause bleaching, corals within the same species can produce either fewer eggs (and maintain size) or smaller eggs (and maintain numbers) depending on their zooxanthellae clade and lipid levels 15 . Corals exposed to elevated nutrients levels also adjusted their reproductive output, with nitrogen reducing both egg size and number of eggs, and phosphorus producing smaller, but more eggs 41 . Furthermore, when coral colonies are transplanted to different latitudes, they adjust their egg size to be similar to local colonies. A transplant study conducted in Taiwan reported that coral colonies transplanted to higher latitudes and cooler waters, developed larger eggs, similar to local colonies, as an increased investment response to unfavourable conditions 36 . This phenotypic plasticity may allow for a type of maternal 'bet-hedging' , where parents increase within clutch variation in offspring phenotype in response to unpredictable environmental conditions 42 . The results of our study showed high within species natural variability, but this variability was not consistent with the trade-off model. That is, while there may have been both large and small mature eggs within a species, the large eggs did not necessarily correspond with fewer eggs in a polyp. Within species size variation amongst offspring has traditionally been underestimated 43 , however, since offspring size can affect dispersal potential, producing a range of sizes, could spread offspring through a range of habitats, thereby spreading the risk of reproductive failure 44 .
It is often assumed that if resources are limited for reproduction, then an increase in egg size should result in a reduction in the number of eggs 37 . However, there are no datasets directly comparing egg size and number within coral species. We have shown that in seven Acropora coral species this trade-off between size and number did not occur. We also did not see any seasonal differences in these measures. We recorded high natural variability in both mature egg size and fecundity, a factor that should not be overlooked when using these measures to gauge or compare reproductive output (e.g. between seasons, years, locations). Since egg size and fecundity are affected by parent colony energy reserves, energy allocation to a range of other life history processes (e.g. growth and repair), polyp size and morphology, responses to environmental conditions, and the interaction of these factors, it is unlikely that there is a simple trade-off between size and number of eggs. It is also unlikely that these measures are constrained only by energetics, given the adaptive phenotypic plasticity reported in other studies 36,40 . Furthermore, parental investment can come in the form of increased egg quality (e.g. lipids or antioxidants), rather than size or number of eggs. More research into coral energetics, natural variability, and adaptive plasticity is required to determine the mechanisms behind some of the patterns we observed, but our study doesn't support a simple trade-off model in coral reproduction.  Figure S1). Field samples were collected from six long-term monitoring locations across the reef system at depths ranging from 3 to 9 m 27 . Key environmental conditions at Scott Reef were summarized to examine seasonal differences in fecundity and egg size.  Figure S2). All seven species were used to determine whether there were relationships between egg size and fecundity. Of these species, some spawn only in autumn (A. humilis, A. polystoma and A. spicifera) and others participate in both spawning seasons (A. gemmifera, A. hyacinthus, A. microclados and A. tenuis) 10 . The four species that spawn during both spawning seasons were used to analyze seasonal differences in egg size and fecundity. Between 15 and 112 (median = 21) colonies per species were sampled in total (Supplementary Table S1). For seasonal comparisons, replication ranged from 5 to 17 (median = 12) colonies per species, per season (Supplementary Table S2).
Laboratory samples. Following collection in the field, Acropora samples were stored in a solution of 90% seawater and 10% formalin. Then in the lab, samples were decalcified in a solution of hydrochloric acid, formaldehyde (37%) and water. The initial solution was 5% HCl and 10% formaldehyde, but with a gradual increase in HCl from 5 to 10% over a period of 2-3 weeks. After decalcification, the tissue samples were stored in 70% ethanol. The tissue samples were then examined to determine the number and size of eggs in each polyp, using a Leica MS205 stereoscope. Five polyps were dissected for each of the three branches sampled per colony. The polyps were sampled from the centre sections of the branch to avoid the sterile growing tips. All of the eggs within each polyp were counted and the maximal diameter measured using Leica Application Suite version 3.1 software. If no eggs were present a further ten polyps were checked to confirm the results.
Data analysis. Data collected in 2008, 2009 and 2010 were used to compare egg counts and egg sizes between the seasons (autumn and spring). Data were checked for normality using the Shapiro-Wilk test and for equality of variance using Levene's test. Two sample t tests were used to compare the means of egg counts per polyp and maximal egg size for each of the species between autumn and spring spawning. Where data did not meet assumptions of normality, data were log transformed (A. hyacinthus egg count data). The correlations between mean maximal egg sizes and counts per polyp within and between species, were determined using regression analyses. To avoid the potentially confounding factor of season, only autumn samples were used for regression analyses in the four biannually spawning coral species. All tests and analyses were conducted in the software R. www.nature.com/scientificreports/