INTRODUCTION
Two energy sources exist for planktotrophic larvae: an endogenous and an exogenous one, with the latter coming from particulate food and dissolved organic matter. Because planktotrophs tend to produce small eggs with relatively low reserves, they should depend more on exogenous than endogenous energy. Indeed, the strong effect of external energy on the growth and development of planktotrophic echinoderms larvae has been extensively documented (Strathmann et al., Reference Strathmann, Fenaux and Strathmann1992; Hart & Strathmann, Reference Hart and Strathmann1994; Basch, Reference Basch1996; Meidel et al., Reference Meidel, Scheibling and Metaxas1999).
Endogenous energy stems from the maternal allocation of nutrients to eggs of planktotrophic larvae. Maternal nutrition, maternal habitat and body size have an influence on the egg size and quality (Jong-Westman et al., Reference Jong-Westman, Qian, March and Carefoot1995; George, Reference George1999). Mothers with a higher nutrition may produce eggs of a larger size and/or with a greater organic content, resulting in higher larval growth, development, survival and metamorphosis rates (George, Reference George1990; George et al., Reference George, Cellario and Fenaux1990; Jong-Westman et al., Reference Jong-Westman, Qian, March and Carefoot1995). However, others believe that the parental nutritional condition and the endogenous energy source have little or no effect on the subsequent characteristics of planktotrophic larvae (McEdward, Reference McEdward1986a; Bertram & Strathmann, Reference Bertram and Strathmann1998; Meidel et al., Reference Meidel, Scheibling and Metaxas1999). Therefore, the importance of parental food rations on larval growth and development remains unclear for planktotrophic echinoderms.
Although a number of studies have examined the effects of parental diet or population on sea urchin larvae (George, Reference George1990; George et al., Reference George, Young and Fenaux1997; Bertram & Strathmann, Reference Bertram and Strathmann1998; Meidel et al., Reference Meidel, Scheibling and Metaxas1999), little is known about these effects on larval features in starfish (see George, Reference George1999).
To this end, we examined the relative contributions of the parental nutritional status and larval nutrition to the growth, development, shape, mortality rate and biochemical composition of Sclerasterias mollis larvae. This species is widely distributed throughout the south-east inshore waters of New Zealand and can be found at depths ranging from 40 to 140 m (Barker & Xu, Reference Barker and Xu1991). Sclerasterias mollis has bipinnaria developing brachiolaria larvae with planktotrophic nutrition. Furthermore, to determine whether variation in larval features can be explained by egg characteristics, the egg diameter, fecundity, fertilization percentage, biochemical and energy content of eggs were examined. Eggs and larvae from both laboratory held and field collected adults were studied because it was not known whether the laboratory provided optimal food and environmental conditions for the adults.
MATERIALS AND METHODS
Collection and conditioning of adult Sclerasterias mollis
Sclerasterias mollis were collected from inshore waters on the Otago continental shelf (New Zealand; 45°47′218″S 170°51′717″E) using an Agassiz bottom trawl and transferred to the laboratory. Three hundred starfish were randomly allocated to six rectangular plastic tanks (82 × 22 × 22 cm) supplied with sand-filtered (50 µm) running seawater (≈2 l min−1) was provided for each tank. Animals were fed a low diet (three tanks, half a cockle Austrovenus stutchburyi with 2–3 cm diameter per starfish every two weeks) or a high diet (three tanks, two cockles per starfish every week), respectively. Tanks were cleaned and scrubbed every two weeks. The experiment was run for one year.
We further sampled two populations with a dissimilar nutritional status from either Otago inshore (45°47′218″S 170°51′717″E; 60 m depth) or offshore (45°48′235″S 170°55′928″E; 100 m depth) waters. The nutritional status of the two populations was inferred through the gonad index, where a higher gonad index indicates better nutrition (Xu & Barker, Reference Xu and Barker1990). The gonad index was measured as the per cent ratio of the gonad wet weight to body wet weight. The Otago inshore S. mollis had significantly greater gonad indices (Otago inshore: N = 96, mean±SE = 5.54±0.57%, Otago offshore: N = 83, 1.71±0.34%; t-test: df = 177, t = 5.525, P < 0.001). When the adults were seasonally ripe (see the section ‘Egg studies’), they were sampled by an Agassiz bottom trawl.
Egg studies
Three months before the beginning of the breeding season (August–September), five S. mollis were selected each month randomly from the tanks or sampled from the Otago inshore and offshore populations. A small piece of the gonad (size 1 cm) was biopsied and inspected under a compound microscope to assess egg maturity. When the oocyte diameters were approaching 100 µm, and both the germinal vesicle and the nucleolus were invisible, nine to ten females were taken from the low and the high diet laboratory groups or sampled from the Otago inshore and offshore populations. The sex of the animals was determined on a fresh slide smear under a compound microscope. Ovaries were excised and put in a 250 ml beaker containing 50 ml of 10−5 M 1-methyladenine. Two hours after the onset of gamete release, ovaries were removed from the solution to avoid a decrease in the fertilization percentage due to eggs sticking to each other. Most ovaries were spent after two hours. A 500 µl sample was taken to measure egg diameter and fecundity. A 5 or 10 ml sample was taken, concentrated in an Eppendorf tube, rinsed with distilled water, and kept at −80°C for later analyses.
The following egg characteristics were measured: (1) egg diameter—the average egg diameter was determined from 30 eggs per female using an ocular micrometer. If eggs were not spherical, the average of the major and minor axes was used as the egg diameter; (2) egg dry weight—a defined number of eggs were put in a dry pre-weighed Eppendorf tube, freeze-dried for 24 hours and weighed; (3) fecundity—the average number of eggs in ten 2 µl subsamples per female; (4) fertilization percentage—for each female, the fertilization percentage of 100 randomly-selected eggs was determined. Inflation of the egg membrane and formation of a perivitelline space was indicative of fertilization.
Protein concentration as well as lipid and carbohydrate content of freeze-dried eggs were measured by the Bradford protein assay (Bradford, Reference Bradford1976) and by colorimetric methods (Mann & Gallager, Reference Mann and Gallager1985), respectively. The egg energy content was estimated using the energy equivalent for protein, lipid and carbohydrate: 5.65, 9.45, and 4.0 kcal g−1, respectively (Brody, Reference Brody1945).
Larval studies
Larvae from each parental treatment were reared at 12°C in six 3 l glass jars at a density of two larvae ml−1 and were kept suspended in water column using a standard paddle system at 10 strokes min−1 (Strathmann, Reference Strathmann1987). Larvae were fed Dunaliella primolecta at densities of 2000 and 8000 cells ml−1 for low and high concentration planktonic diets, respectively. The experimental design was full-factorial (4 parents: low and high diet laboratory parents, Otago inshore and offshore field parents × planktonic diet concentrations ×3 jars). Algae cultures were replaced every two weeks. Water levels were lowered every four days by 90% and replenished with fresh 1 µm filtered seawater. At the same time, the culture jars were cleaned. Larviculture lasted for 300 days.
Every 15 days, five intact larvae were sampled randomly from each jar. The developmental stage was determined according to Basch (Reference Basch1996) and classified as: (1) early-bipinnaria 1—coelom lower to mouth; (2) early-bipinnaria 2—coelom parallel to mouth; (3) early-bipinnaria 3—coelom upper to mouth and unfused; (4) early mid-bipinnaria—coelom upper to mouth and fused; (5) mid-bipinnaria—coelom extended to anterior process; (6) early-brachiolaria—brachiolar arm buds appear; and (7) brachiolaria—presence of adhesive disc and brachiolarian arms. Body length, body width, mouth width, oesophagus length, coelomic pouch length, stomach length and width (Figure 1) were measured using an ocular micrometer.
Morphometric dimensions measured in a Sclerasterias mollis bipinnaria larva. (A) Body length; (B) body width; (C) mouth width; (D) oesophagus length; (E) stomach length; (F) stomach width; (G) coelomic pouch length.
Every 15 days, three 5 ml samples were taken from the middle of each jar and the number of larvae was counted using a Bogorov counting tray under a dissecting microscope before being returned to the jars. Larval density in the jars were derived from the average of triplicate measurements at a specific sampling time. Only data for the first 90 days were used for the analysis, because thereafter larval density decreased to levels not measurable by this technique.
Larvae from each parental treatment were reared at a larval density of 2 ml−1 in two 67 l white plastic tanks (0.53 m height × 0.51 m diameter) at 12–14°C, with one tank for the low and another for the high concentration planktonic diet. Tanks were equipped with paddles connected serially to a gearbox motor (1/8 HP) generating 12.5 strokes per minute. A 50 µm mesh round filter (10 cm diameter) protected the outlet on the top of each tank. The diet was as described above for the jars. Water was changed every two days by adding 30 l of 1 µm filtered isothermal seawater to each tank. Tanks were discharged, scrubbed and washed with fresh water every two weeks. Every 16 days, three samples of 3000 larvae each were taken, concentrated in Eppendorf tubes, rinsed with distilled water and kept at −80°C for later biochemical analyses.
Data analysis
EGG STUDIES
Significant differences among the egg characteristics from laboratory or field parents were examined using a one-way ANOVA. Normality and homogeneity of variance were tested using the Shapiro–Wilk and the Levene tests, respectively.
LARVAL STUDIES
Principal component analysis (PCA) was used to determine differences among parents and larval planktonic diets in larval growth (size-at-age) and shape (the relative size of the body length-to-width). Body length, body width, mouth width, oesophagus length, stomach length, stomach width and coelomic pouch length were analysed. To avoid pseudoreplication, the average of each morphometric variable was calculated over the jar prior to analysis (Bertram & Strathmann, Reference Bertram and Strathmann1998). The first two principal components (PC) were chosen based on a scree diagram (Everitt & Dunn, Reference Everitt and Dunn1991). A correlation matrix was used for PCA as the variables had very different variances (Quinn & Keough, Reference Quinn and Keough2002). Because all variables were strongly skewed, a logarithmic transformation was applied to improved normality. PC1 was considered as a growth component. PC2 was considered as a shape component, because the body length coefficient contrasted with the body width coefficient. Therefore, larger scores for PC2 would show a smaller body length relative to the body width. To test the difference in growth and shape between larvae from different parents and larval planktonic diets, the means of the PC scores were compared using Wilcox's three-way design (Wilcox, Reference Wilcox2005), with parents, larval diet and age as fixed factors. This test is based on 20% trimmed means and the bootstrap method. ANOVA could not be applied, because the assumptions for data normality and variance homogeneity were not given.
The instantaneous mortality rate of the larvae was calculated using the exponential model (George et al., Reference George, Cellario and Fenaux1990):
where Nt = number of larvae alive at time t (days), N0 = number of larvae at time t = 0 (2 ml−1), e = the base of the natural logarithm (≈2.71) and K = instantaneous mortality rate (day−1).
The effects of parents, larval diet, and age on larval mortality rates, protein, lipid, and carbohydrate concentrations were examined by the three-way Wilcox test, because it was not possible to meet the ANOVA assumptions. Parents, larval diet and age were fixed factors.
Main effects in ANOVA and the Wilcox three-way design were addressed if no interaction between factors was observed. If a significant interaction between parents and larval diet was present, levels of one factor were compared with each level of the other factor (Underwood, Reference Underwood1997) using a Tukey or t-test. A significant interaction among parents, larval diet and age was interpreted as parents × larval diet and age (Underwood, Reference Underwood1997). In other words, levels of parents were compared within each level of larval diet (and vice versa) at each larval age. The family-wise probability of Type-I error was adjusted using the Bonferroni method (0.05 divided by the number of multiple comparisons; Underwood, Reference Underwood1997).
Multinomial logistic regression and likelihood ratio tests were used to determine if parents and larval diet had a significant effect on larval development. Due to singularities in the Hessian matrix, jar and larval age were omitted sequentially. Eliminating jar and larval age did not affect the parents and larval diet effects. Only main effects were tested, because the inclusion of the interaction of the parents and the larval diet in the model would decrease the degrees of freedom of main effects to zero. Ordinal regression was not used, as the test of parallelism revealed that the relationships between the independent variables and the logits were not the same for the all logits (Norusis, Reference Norusis2008). Statistical analyses were performed using SPSS 15.0 (release 15.0.0) and R (version 2.7.1, Rand R. Wilcox's allfun package).
RESULTS
Egg studies
Eggs from the low diet and the Otago offshore parents were larger than those from the high diet and the inshore parents (Table 1). Eggs from parents collected in the field had a higher fertilization rate but a lower carbohydrate concentration than eggs from parents held in the laboratory (Table 1). Thus, restricted diet is associated with an increased egg size.
Egg characteristics and one-way ANOVA for the effect of Sclerasterias molli s held in the laboratory or collected from Otago inshore and Otago offshore on egg features. Sample statistics are mean±SE with sample size in parentheses. Carbo., carbohydrate. Significant P values are shown in bold. Lack of at least one similar letter between parents indicates a significant difference (α = 0.05) detected by a Tukey test (B > A).
Larval growth and shape
EFFECTS OF PARENTS
The PC1 coefficients were positive for all larval body components and indicated body growth, whilst the PC2 coefficients for the body length contrasted with those for the body width and thus indicated the shape of the larvae (Table 2).
Principal component coefficients and eigen values generated by PCA for Sclerasterias molli s larvae from the parents held with a low or high diet in the laboratory, and from Otago inshore and offshore parents.
All larval body components increased significantly over time and reached sizes greater than those at the beginning of the experiment (Figures 2 & 3; Table 3). However, in larvae derived from the laboratory held parents and fed a low concentration planktonic diet, the mouth width, stomach length and stomach width did not grow after day 45.
Mean (±SE) body components of Sclerasterias mollis larvae from low (○) and high (•) diet laboratory parents under low and high concentrations planktonic diets. Means were calculated from 15 larvae from three jars.
Mean (±SE) body components of Sclerasterias mollis larvae from Otago inshore (□) and Otago offshore (■) field parents under low and high concentrations planktonic diets. Means were calculated from 15 larvae from three jars.
Three-way Wilcox test for the effects of parents, larval diet and age on the first and second principal component scores of Sclerasterias molli s larvae. Significant P values are shown in bold.
PC1 and PC2 scores of larvae from the laboratory and the field parents are shown in Figure 4. There was a significant interaction among parents, larval diet and age in PC1 scores (Table 3). Tukey tests between levels of parents (low and high diets, Otago inshore and offshore populations) indicated that, under a low concentration planktonic diet, larvae from Otago offshore parents had greater mean PC1 scores than those from other parents at days 30, 210, 225, 255, 270 and 285. Under a high concentration planktonic diet, larvae from Otago offshore parents had higher mean PC1 scores at days 30 and 240, indicating their larger body components than larvae from other parents. There was no significant difference in the mean PC1 scores between larvae from low and high diet laboratory parents. Hence, larvae from Otago offshore parents had the largest body components.
Means of the first and second principal components scores of Sclerasterias mollis larvae from low (○) and high (•) diet laboratory parents and from Otago inshore (□) and Otago offshore (■) field parents under a low or high concentration planktonic diet. Data are pooled over jar and larval age.
There were significant interactions among parents, larval diet and age in the mean PC2 scores (Table 3). Further Tukey tests between levels of parents indicated that, under a low concentration planktonic diet, larvae from Otago offshore parents had larger mean PC2 scores at days 60, 105, 210, 225, 255 and 285 than larvae from other parents. Under a high concentration planktonic diet, larvae from Otago offshore parents had higher mean PC2 scores at day 240, indicating their smaller body length relative to the body width than larvae from other parents. There was no significant difference in the mean PC2 scores between larvae from low and from high diet laboratory parents. Thus, larvae from Otago offshore parents showed the highest rate of morphological phenotypic plasticity.
EFFECTS OF LARVAL DIET
PC1 scores indicated a significant interaction among parents, larval diet and age in mean PC1 scores (Table 3). A t-test was used to compare the PC1 scores from larvae fed low or high concentration planktonic diets within each level of parents and at each larval age; no significant difference was observed between low and high concentration planktonic diets in the PC1 scores from larvae derived from low-diet, laboratory-held parents. Larvae from high-diet, laboratory-held parents and fed a high concentration planktonic diet had greater mean PC1 scores at days 240 and 285 than when fed a low concentration planktonic diet. Larvae from Otago inshore parents and fed a high concentration planktonic diet had greater mean PC1 scores than low diet larvae at day 30. Larvae from Otago offshore parents and fed a high concentration planktonic diet had a greater mean PC1 score at days 45 and 60, indicating larger body components than those of larvae fed a low concentration planktonic diet at these days. Therefore, concentration of planktonic diet had a short-term effect on the size of larval body components.
PC2 scores indicated a significant interaction among parents, larval diet and age in the mean PC2 scores (Table 3). Using a t-test, PC2 scores from larvae fed low or high concentration planktonic diets were compared within each level of parents and at each larval age. Larvae from low-diet laboratory parents and fed a high concentration planktonic diet had greater mean PC2 scores at day 165 than when fed the ‘low’ diet. Larvae from high-diet laboratory parents and fed a high concentration planktonic diet had greater mean PC2 scores at days 240, 255 and 300. Larvae from Otago offshore parents and fed a high concentration planktonic diet had significantly greater mean PC2 scores at days 15 and 285, indicating a smaller body length-to-body width-ratio than that of the ‘low’ diet group. Larval diet had no significant effect on mean PC2 scores of larvae from Otago inshore parents. Hence, concentration of planktonic diet had a short-term effect on shape of the larvae.
Based on the Wilcox test Q statistic, parents had a stronger impact on larval growth (based on PC1 scores) and shape (based on PC2 scores; Table 3) than the planktonic diet.
Larval development
Parents had a significant effect on the larval development rate (−2 log likelihood of reduced model: 430.600; likelihood ratio tests: df = 21, χ2 = 156.072, P < 0.001). Larvae from parents held in the laboratory on a low diet and those from Otago inshore parents had a faster development than the larvae from high diet laboratory and Otago offshore parents, respectively (compare Figure 5A with C and B with D).
Percentage of Sclerasterias mollis larvae in developmental stages from low and high diet laboratory parents and from Otago inshore and offshore field parents under a low or high concentration planktonic diet. Each bar is calculated from measurement of 15 larvae. The developmental stages are early-bipinnaria 1: 
, early-bipinnaria 2: 
, early-bipinnaria 3: 
, early mid-bipinnaria: 
, mid-bipinnaria: 
, early-brachiolaria: 
and brachiolaria: 
.
Larval diet had a significant effect on the larval development rate, with larvae on a high planktonic diet having a faster development rate (–2 log likelihood of reduced model = 446.466, likelihood ratio tests: df = 7, χ2 = 171.939, P < 0.001; compare Figure 5A with B, C with D, E with F, and G with H).
Larval mortality
Neither parental nor nutritional conditions had a significant effect on larval mortality (Table 4; Figure 6). The only parameter significantly affecting the larval mortality rate was time.
Mean (±SE) instantaneous mortality rate of the larvae from low (○) and high (•) diet laboratory parents and from Otago inshore (□) and offshore (■) parents under a low or high concentration planktonic diet. Means are calculated from three measurements.
Three-way Wilcox test for the effects of parents, larval diet and age on mortality rate of Sclerasterias mollis larvae. Significant P value is shown in bold.
Biochemical composition of the larvae
EFFECTS OF PARENTS
Parents had no significant effect on protein and lipid concentration of the larvae (Table 5). A three-way Wilcox test detected a significant interaction between parents and larval diet with regards to the larval lipid concentration (Table 5). However, further Tukey and t-tests at each larval age indicated that parents and larval diet had no significant effect on the lipid concentration (Figure 7).
Mean (±SE) of biochemical composition of the larvae from low (○) and high (•) diet laboratory parents and from Otago inshore (□) and offshore (■) parents under a low or high concentration planktonic diet. Means are calculated from three measurements.
Three-way Wilcox test for the effect of parents, larval diet and age on protein, lipid and carbohydrate concentrations of Sclerasterias molli s larvae. Significant P values are shown in bold.
There was a significant interaction among parents, larval diet and age regarding the larval carbohydrate concentration (Table 5). Thus, further Tukey and t-tests comparing parental levels within each level of the larval diet and at each larval age were performed. Larvae from low-diet, laboratory-held parents had a greater mean carbohydrate concentration than the larvae from other parents at days 8 and 86 (Figure 7). Hence, parent only had a significant effect on carbohydrate concentration of the larvae.
EFFECTS OF LARVAL DIET
Larval diet had no significant effect on the protein and lipid concentration of larvae (Table 5). There was a significant interaction among parents, larval diet and age regarding the larval carbohydrate concentration (Table 5). Further Tukey and t-tests comparing the various larval groups revealed that larvae from low-diet, laboratory-held parents and fed a high concentration planktonic diet had a greater mean carbohydrate concentration than the larvae on a low planktonic diet at day 86. Hence, concentration of planktonic diet had a short-term significant effect on carbohydrate concentration of larvae.
DISCUSSION
Egg studies
Parents with a lower nutritional status (i.e. low diet in the laboratory or from Otago offshore) produced larger eggs than those with a high nutritional status. Compared to laboratory-held parents, the field starfish produced eggs with a higher fertilization rate and a lower carbohydrate concentration.
Most life-history models have interpreted changes of the egg size in relation to the egg number (Vance, Reference Vance1973a, Reference Vanceb; McEdward, Reference McEdward1997). These models predict that in favourable conditions females produce a large number of small eggs but few large eggs in unfavourable conditions. However, these models cannot explain the larger eggs from parents with a lower nutritional status as observed in this study, because egg number was not inversely related to the egg size. Furthermore, it has frequently been demonstrated that the relationship between the egg size and number deviates from what models on reproductive strategy have predicted. For example, George (Reference George1994) found that the starfish Leptasterias epichlora from a site with more resources produced a higher number of larger eggs than those from a resource-poor site. Additionally, in spite of many existing studies, it is difficult to draw a general inference on the topic. For instance, Thompson (Reference Thompson1982), George et al. (Reference George, Lawrence and Fenaux1991) and George (Reference George1999) found larger eggs in food-limiting echinoderms, while George et al. (Reference George, Cellario and Fenaux1990) and George (Reference George1990) found larger eggs in populations that seemed to have better nutritional status. Thus, changes of the egg size corresponding to nutritional status may be a species-specific trait and further studies are suggested.
Field and laboratory parents had different effects on the egg diameter and carbohydrate concentration. Two potential scenarios may explain this difference: firstly, when the mean egg diameter of starfish fed a high diet was compared to that of starfish fed a low diet, the latter tended to be larger, albeit at non-significant levels. Meidel & Scheibling (Reference Meidel and Scheibling1999) have shown that the duration of feeding experiments influences oocyte size in Strongylocentrotus droebachiensis. They found that females fed low and high diets had a similar oocyte size at first reproduction but a significantly different oocyte size at second reproduction. Hence, in the present study, the food-dependent changes in Sclerasterias mollis egg dimensions may take a longer time to develop and may become significant in the following years. Secondly, the biochemical egg composition is influenced by parental diet (Thompson, Reference Thompson1982; Jong-Westman et al., Reference Jong-Westman, Qian, March and Carefoot1995). Therefore, the higher egg carbohydrate concentration in laboratory held parents may suggest that laboratory and field diets were not similar in the current study.
Larval growth, development and shape
Larval nutrition had a significant influence on the larval development rate, growth and shape. Larvae fed a high concentration planktonic diet had larger body components, a faster development rate and smaller body length relative to the body width at some sampling times. The influence of planktonic diet on growth and shape was however limited to a few sampling times. These results were expected as Sclerasterias mollis larvae have a planktotrophic development, and an exogenous diet is their main energy source. These findings are consistent with previous studies on other starfish larvae such as Luidia clathrata (George et al., Reference George, Lawrence and Fenaux1991), Pisaster ochraceus (George, Reference George1999) and Asterina miniata (Basch, Reference Basch1996). In this study, the limited effect of planktonic diet on larval growth and development suggests that Dunaliella primolecta was not the preferred diet for S. mollis larvae. This suggestion is further supported by the fact that the larvae had a slow development rate under both low and high concentration planktonic diets, resulting in a long period to reach the brachiolaria stage.
The larval growth, shape and development rate were different between laboratory and field parents. Field parents but not laboratory held parents had a significant influence on larval growth and shape, with larvae from Otago offshore parents having a stronger growth and a smaller body length relative to the body width than larvae from other parents. In addition, both larvae from low diet laboratory parents and from Otago inshore parents (assumed to have higher nutritional status) had a fast development rate. The inconsistencies in larval growth, shape and development rates between laboratory and field parents suggest that the difference in the diet of the low and high fed laboratory parents was not severe enough to affect larval characteristics. In addition, other parameters (e.g. depth and temperature) which could not be assessed with the present study design may have contributed to these differences. However, most echinoderm studies on this topic have investigated mainly the parental nutritional status as a factor affecting larval features (George, Reference George1990, Reference George1994, Reference George1996, Reference George1999; Bertram & Strathmann, Reference Bertram and Strathmann1998). In this respect, the importance of non-nutritional factors (e.g. depth and temperature) has been widely neglected in echinoderm research and needs to be addressed in future studies.
Parents and larval diet had significant effects on the larval shape. Phenotypic plasticity in response to varying concentrations of planktonic diet has been reported previously (George, Reference George1999). Nonetheless, the present study is the first to report a significant effect of parents on morphological phenotypic plasticity of starfish larvae.
Echinoderms with a better nutritional status may produce larger eggs and/or eggs with a greater amount of biochemical reserves (protein, lipid and carbohydrate) and energy, which may subsequently result in stronger larval growth, development, survival and metamorphosis rates (George, Reference George1990; George et al., Reference George, Cellario and Fenaux1990, 1991; Jong-Westman et al., Reference Jong-Westman, Qian, March and Carefoot1995). In the current study, Otago offshore parents produced the largest eggs and their larvae had the best growth and greatest morphological phenotypic plasticity. Thus, our results appear to agree with previous studies suggesting that parents affect larvae through egg features. However, it is unlikely that egg size is the only factor affecting larval growth and phenotypic plasticity because Otago inshore parents had smaller eggs than Otago offshore parents but their larvae had a faster development rate. Furthermore, the faster development rate of the larvae from both low diet laboratory and Otago inshore parents cannot be explained by egg reserves and energy, because their egg sizes were similar. Therefore, effects of starfish parents on their larvae are not exerted solely through the size and the biochemical content/energy of the eggs.
Larval mortality and biochemical composition
Neither parents nor planktonic diet concentrations had a significant effect on the larval mortality rate. The effect of parents on larval mortality rates has been related to egg resources (Jong-Westman et al., Reference Jong-Westman, Qian, March and Carefoot1995; George, Reference George1999) and feeding efficiency (George et al., Reference George, Cellario and Fenaux1990) in echinoderms. In the present study, the resources allocated to eggs (the biochemical and energy content of the egg) were not influenced by the parental nutritional status. However, there was a difference in phenotypic plasticity of the larvae from laboratory and field parents (see above, section ‘larval growth, development and shape’), which might be related to shape changes that ultimately affect feeding efficiency (McEdward, Reference McEdward1986a, Reference McEdwardb; George, Reference George1999). However, the similar mortality rates of the larvae from laboratory and field parents in this study are unlikely to be related to the feeding efficiency, because the concentration of planktonic diet had no influence on larval mortality.
It is believed that food availability is not a serious problem for the survival of marine invertebrate larvae (Manahan, Reference Manahan1989; Moran & Manahan, Reference Moran and Manahan2004), even though a low planktonic diet may prolong the larval period and thus increase the risk of mortality by non-nutritional factors (Morgan, Reference Morgan and McEdward1995). The results of the present study are consistent with this view, given that ‘low diet’ larvae had a slower development rate and consequently a longer larval period than ‘high diet’ larvae. However, the planktonic diet concentration had no effect on the mortality rate of the larvae.
Parents and larval diet had a significant effect on the larval carbohydrate concentration. These results were expected, as low-diet, laboratory-held parents that produced eggs with the highest carbohydrate concentration also had larvae with the greatest carbohydrate content. This result is consistent with studies by George (Reference George1990, Reference George1994) who assigned the biochemical composition of the larvae to the egg reserves, and with studies by Schiopu et al. (Reference Schiopu, George and Castell2006) and George et al. (Reference George, Fox and Wakeham2008) who demonstrated that larval diet influences larval composition.
In summary, this study indicates that: (i) parents have a significant effect on the characteristics of Sclerasterias mollis larvae; and (ii) parents can be more important than larval diet with respect to some larval features (e.g. growth, shape and biochemical composition). The influence of the parents on S. mollis larvae cannot be simply explained by egg characteristics such as size, organic content and energy. Indeed, biochemical egg components other than protein, lipid and carbohydrate can significantly affect the development rate of echinoderm larvae. For example, Saito et al. (Reference Saito, Seki, Amemiya, Yamasu, Suyemitsu and Ishihara1998) have observed that the levels of thyroid hormones in the egg influence the larval development rate in the sea urchin Peronella japonica. Further, egg carotenoids have also been associated with the larval development in the sea urchins Heliocidaris tuberculata and H. erythrogramma (Tsushima et al., Reference Tsushima, Byrne, Amemiya and Matsuno1995). Therefore, our study is consistent with the presence of uncharacterized egg compounds that mediate the parental effects on the characteristics of S. mollis larvae, and suggests searching for these factors in future studies.
ACKNOWLEDGEMENTS
The authors appreciate the help of staff of the Portobello Marine Laboratory, Dr Bostjan Humar for kindly reading and improving the manuscript and the Iranian ministry of science for a PhD scholarship and a Postgraduate Publishing Bursary from the University of Otago to Hadi Poorbagher.
