Marine Biology

, Volume 153, Issue 3, pp 337–349

Fuels for development: evolution of maternal provisioning in asterinid sea stars

Authors

    • Discipline of Anatomy and Histology, Bosch Institute, F13University of Sydney
  • M. A. Sewell
    • School of Biological SciencesUniversity of Auckland
  • M. Byrne
    • Discipline of Anatomy and Histology, Bosch Institute, F13University of Sydney
Research Article

DOI: 10.1007/s00227-007-0809-7

Cite this article as:
Prowse, T.A.A., Sewell, M.A. & Byrne, M. Mar Biol (2008) 153: 337. doi:10.1007/s00227-007-0809-7

Abstract

For marine invertebrates, larval developmental mode is inseparably linked to the nutritional content of the egg. Within the asterinid family of sea stars there have been multiple, independent, evolutionary transitions to lecithotrophic development from the ancestral, planktotrophic state. To investigate the evolution of maternal investment and development within the Asterinidae, we quantified individual lipid classes and total protein for eggs and larval stages of closely related species representing three developmental modes (planktotrophy, planktonic lecithotrophy and benthic lecithotrophy). Within species, maternal provisioning differed between females indicating that egg quality varied with parentage. Maternal investment was related to egg size but, after correcting for egg volume, we identified two major oogenic modifications associated with the evolution of lecithotrophic development: (1) a reduction in protein deposition that probably reflects the reduced structural requirements of nonfeeding larvae, (2) an increase in deposition of a single class of energetic lipid, triglyceride (TG). The exception was Parvulastra exigua, which has benthic, lecithotrophic development and lays eggs with a lipid to protein ratio close to that of planktotrophs. This oogenic strategy may provide P. exigua larvae with a protein “weight-belt” that assists in maintaining a benthic existence. Asterinids with planktotrophic development used a significant portion of egg TG to build a feeding bipinnaria larva. For Meridiastra mortenseni, female-specific differences in egg TG were still evident at the bipinnaria stage indicating that egg quality has flow-on effects for larval fitness. In lecithotrophic asterinids, TG reserves were not depleted in development to the larval stage whereas protein stores may help fuel early larval development. Available data indicate that there may be two evolutionarily stable egg lipid profiles for free-spawning, temperate echinoderms.

Introduction

For marine invertebrates, developmental mode and larval ecology are inseparably linked to egg size and nutritive content (Emlet et al. 1987). Species with small eggs typically have a long dispersive phase through a feeding (planktotrophic) larva that must accumulate energetic reserves prior to metamorphosis by feeding in the plankton. In contrast, species that spawn large eggs usually have a short dispersive phase through a nonfeeding (lecithotrophic) larva. With the notable exception of some spiralian taxa (Kupriyanova 2003; Collin 2004), planktotrophy is believed to be the ancestral state from which lecithotrophic development has evolved in most marine invertebrates (e.g. Strathmann 1985; Duda and Palumbi 1999; McEdward and Miner 2001). The regaining of a feeding larva after the evolution of nonfeeding development appears far less likely (Wray 1995; Hart 2002).

The Echinodermata have proven extremely useful for the study of developmental evolution due to the presence of developmental diversity among closely related species (Raff and Byrne 2006). Evolution of lecithotrophy in echinoderms has involved an increase in maternal investment per egg and is associated with bimodal egg size distributions in some classes (Sewell and Young 1997). Such egg size dichotomies reflect two strategies of maternal investment associated with obligate feeding and nonfeeding development. Echinoderms with intermediate sized eggs and facultatively feeding larvae (i.e. larvae that are able to feed but do not need to feed to complete metamorphosis) probably represent a transitional state between planktotrophy and lecithotrophy (Emlet 1986; Wray 1996). There has therefore been much interest in the potential selective pressures underlying increases in maternal investment in this phylum. A plethora of studies has investigated the effect of egg size on fertilisation kinetics, the length of the facultative feeding period, phenotypic plasticity, larval size, developmental rate, metamorphic success and juvenile size and performance (e.g. Levitan 1993; Emlet and Hoegh-Guldberg 1997; George 1999; Miner et al. 2005; Styan et al. 2005; McEdward and Miner 2006).

Although egg size is a good indicator of maternal investment (Jaeckle 1995; McEdward and Morgan 2001) we currently lack a comprehensive understanding of the nutritional requirements of embryos and larvae of planktotrophic and lecithotrophic developers. Energetic content of echinoderm eggs has previously been estimated by the ash free dry weight or dichromate oxidation methods (e.g. McEdward et al. 1988; McEdward and Chia 1991; Moreno and Hoegh-Guldberg 1999) but these approaches provide little detail and the latter method is inaccurate for small eggs (Pernet and Jaeckle 2004). Assimilation of data for total egg lipid, protein and carbohydrate content provides a reliable estimate of maternal investment (e.g. Raymond et al. 2004; Reitzel et al. 2005) but a colorimetric or gravimetric measure of total lipid seems inadequate now that individual lipid classes are easily quantified (Ackman 1999).

Lipid compounds play structural (e.g. cholesterol, phospholipid) and energetic (e.g. triglyceride, wax ester) roles in animals. Eggs of echinoderms with lecithotrophic development are lipid-rich in comparison to those of planktotrophs (e.g. Jaeckle 1995; Byrne et al. 1999a, b; Falkner et al. 2006). This leads to two questions: (1) what is the nature of the additional lipid invested in the eggs of lecithotrophic developers? (2) how are these extra provisions used during development? The asterinid family of sea stars provides an excellent model for examining these questions.

Within the Asterinidae there have been multiple, parallel evolutions of a large egg and lecithotrophic larvae from the ancestral planktotrophic state (Byrne 2006). Comparative morphology and histochemistry reveal marked differences between eggs of planktotrophic and lecithotrophic asterinids and demonstrate that even within lecithotrophic species not all eggs are created equal (Byrne et al. 1999a, 2003). Asterinids with nonfeeding development produce large eggs with a high lipid content and a range of buoyancies from negatively buoyant “benthic” eggs to highly buoyant pelagic eggs (Villinski et al. 2002; Byrne 2006). The study of Villinski et al. (2002) based on thin-layer chromatography indicated that eggs of a planktotrophic asterinid contained triglyceride while those of several species with nonfeeding development produced eggs dominated by wax ester.

In this study, we assessed the hypothesis that evolution of nonfeeding development in the Asterinidae was associated with a switch to wax ester as the primary maternally derived energetic lipid. To do so, we extended the phylogenetic sampling of Villinski et al. (2002) and examined maternal investment in six asterinid species from three genera for which phylogenetic relationships have been established (Byrne 2006). Four of these species are in the Meridiastra clade where molecular phylogeny provides strong support that the planktotrophic developer (Meridiastra mortenseni) is in a position basal to three species with pelagic lecithotrophic development (Meridiastra oriens, Meridiastra calcar and Meridiastra gunnii). Also included were species from two sister clades, the planktotroph Patiriella regularis and the benthic lecithotroph Parvulastra exigua.

The planktotrophic Patiriella regularis and M. mortenseni are sympatric on New Zealand rocky shores and were previously considered to be conspecific (O’Loughlin et al. 2002; O’Loughlin and Waters 2004). Larval development of P. regularis (9–10 weeks) has been documented (Byrne and Barker 1991) but this study is the first to examine development of M. mortenseni. M. oriens, M. calcar and M. gunnii occur along the temperate Australian coastline and develop through pelagic, nonfeeding, brachiolaria larvae in 3–4 weeks. Parvulastra exigua, also a temperate Australian species, deposits sticky egg masses on rock surfaces in intertidal pools and develops through a highly modified, benthic brachiolaria larva in 12–14 days (Byrne 1995).

We focused on egg lipid and protein content because these compounds make up approximately 80% of the total energy available in echinoderm eggs (Jaeckle 1995). Details of maternal provisioning in these asterinids were characterised to assess the changes associated with egg size evolution. The small eggs of the planktotrophic developers were considered to represent the ancestral-type asterinid egg. We examined the scaling of egg lipid classes and protein content with egg size and considered how egg lipid to protein ratios are related to location of development (pelagic versus benthic). Given that the energetic cost of constructing feeding and nonfeeding larvae is expected to differ, we documented utilisation of egg nutrients to development of the larval stage for two planktotrophic and two lecithotrophic species. By considering eggs and larvae from individual females, we investigated whether maternal provisioning and utilisation of lipid and protein varied with parentage. Finally, available data on egg lipid classes for temperate echinoderms were used to examine evolution of egg lipid profiles in this phylum.

Materials and methods

Egg and larval sampling

In 2004–2005, M. oriens, M. calcar and Parvulastra exigua were collected in Sydney, New South Wales, from either Clovelly (33°54.52′S, 151°16.03′E) or Gordon’s Bay (33°54.58′S, 151°15.56′E) during their reproductive season (Byrne 1992). M. gunnii was collected in Adelaide, South Australia, from Seacliff Rocks (35°02.17′S, 138°30.44′E) in October 2005. M. mortenseni was sourced in Auckland, New Zealand, from Mission Bay (36°50.47′S, 174°50.10′E) in December 2005. Patiriella regularis was collected from two locations: (1) Matheson’s Bay, New Zealand (36°18.10′S, 174°48.65′E) in February 2004, (2) Hobart, Tasmania (42°53.15′S, 147°20.20′E) in November 2005.

Sea star eggs were obtained in one of two ways: (1) ovaries were excised and placed in 10−5 M 1-methyladenine (1-MA) in 1-μm filtered sea water (FSW) until eggs were released, or (2) females were immersed in FSW containing 3 × 10−5 M 1-MA until spawning occurred. Eggs were rinsed three times with FSW and egg diameters (d) measured using an ocular micrometer or determined from digital photographs using image analysis software (ImageJ, NIH). Egg volume (EV) was calculated using the formula: EV = (1/6)πd3. Five females were sampled per species and replicate egg samples from each female were collected in microcentrifuge tubes for lipid (n = 3) and protein (n = 3) analysis. Sample tubes were centrifuged briefly, excess sea water removed and samples stored at −80°C until analysis. Egg numbers placed in each tube ranged from 10–30 to 500–700 per replicate for lipid analysis and from 30–50 to 2,000 for protein analysis, for species with lecithotrophic and planktotrophic development, respectively.

The amount of lipid and protein used to form a feeding bipinnaria or nonfeeding brachiolaria larva was documented for Patiriella regularis, M. mortenseni, M. calcar and Parvulastra exigua. Although P. regularis females were sampled from two sites, all females used to start larval cultures were sampled from Hobart, Tasmania. For all species except P. exigua, eggs were obtained from three females. Eggs from each female were split into three batches and lipid and protein samples were collected from each batch and stored as detailed above. The remaining eggs were fertilised with a dilute solution of sperm pooled from three males. Each batch of fertilised eggs was used to generate two larval cultures (one each for protein and lipid analyses) resulting in six cultures per female. Due to the low fecundity of P. exigua, three “groups” of eggs were derived from multiple females and handled as for a single female above. Cultures were maintained in 1-μm FSW in a constant temperature room at 18–19°C and water changes conducted after hatching and every day thereafter.

Embryos of the planktotrophs Patiriella regularis and M. mortenseni were reared in replicate 600 ml glass beakers at initial densities of 15 ml−1. For biochemical analyses, cultures were repeatedly sampled at the following stages: hatched gastrula (1 day), early bipinnaria (2 days) and bipinnaria (3 days). These larvae were competent to feed three days post-fertilisation, so samples at 3 days were fully developed feeding larva (Byrne and Barker 1991). For the lecithotrophic M. calcar and Parvulastra exigua, embryos were reared in replicate 400 ml beakers at an initial density of 1.5 ml−1 and sampled at the following stages: blastula (20 h), late gastrula (2 days) and brachiolaria (4 days for M. calcar, 5 days for P. exigua, Byrne 1995).

Lipid extraction and analyses

Lipid was extracted from frozen egg, embryo and larval samples using the procedure of Sewell (2005) with minor modifications. Briefly, each sample was homogenised in ultra-pure water (250 μl) via sonication and transferred to a glass V-vial (Wheaton) along with 25 μl of a ketone internal standard in chloroform. Methanol (250 μl) and chloroform (100 μl) were added and the vial shaken vigorously for 2 min before centrifugation at 1,000 rpm for 5 min. Both the aqueous and chloroform fractions were transferred to a second V-vial, leaving the solid non-lipid material behind. Water (250 μl) and chloroform (250 μl) were then added, the vial was shaken and centrifuged as before, and the lower chloroform layer was retained. This layer contained the sample lipids and was dried down in a stream of N2 gas and redissolved in a known volume of chloroform (10–20 μl).

Lipid classes were identified and quantified using an Iatroscan Mark Vnew Thin-Layer Chromatography/Flame Ionization Detection (TLC/FID) system and the protocols of Parrish (1999) with minor modifications. In short, 1 μl of each sample lipid solution was spotted on to replicate silica Chromarods (n = 2) and lipid classes were clearly separated chromatographically using a triple-development system (Parrish 1999). After each development, a portion of every Chromarod was analysed via hydrogen flame ionisation detection, resulting in three chromatograms per rod. Quantification of lipids was achieved using calibration curves produced for each lipid class on a rack of ten Chromarods. Lipid standards used were the same as in Sewell (2005) and represented aliphatic hydrocarbon (AH), wax ester (WE), triglyceride (TG), free fatty acid (FFA), alcohol (ALC), sterol (ST), diglyceride (DG), acetone mobile polar lipid (AMPL) and phospholipid (PL). Methyl dodecanoate (Sigma) was used as a representative methyl ester (ME) (Parrish 1987).

Total soluble protein extraction and analyses

Frozen samples for protein analysis were sonicated on ice for 20 s in 100–300 μl of homogenisation buffer (20 mM Tris-HCl (pH 7.6), 130 mM NaCl, 5 mM EDTA) containing 10 μl/ml of both Triton-X and Protease Inhibitor Cocktail (Sigma). The sample homogenate was shaken on ice for 15 min, centrifuged at 4°C for 20 min at 18,000×g, and the supernatant retained. Total soluble protein was measured at 550 nm using a Micro BCA™ Protein Assay Kit (Pierce) and a microplate reader (BioRad).

Statistical analyses

Prior to analysis, the assumption of homogeneity of group variances was examined using residual analysis (Quinn and Keough 2002). Data were log transformed when necessary to improve the spread of the residuals. P. regularis females were collected from two sites (see above) but this had no effect on egg lipid or protein content (unbalanced nested ANOVAs, F(1,3) = 1.05 and 0.35 respectively, both P > 0.25) so collection site was not considered as a factor in subsequent analyses.

We wished to control the experimentwise error associated with multiple analyses of variance (ANOVA) without excessive loss of power and therefore used Dunn-Sidak transformed Type I error rates (Quinn and Keough 2002) calculated independently for 3 families of data: (1) egg data (eight tests, α = 0.006), (2) TG content during development (four tests, α = 0.013), (3) protein content during development (four tests, α = 0.013). Although multiple terms were tested in each ANOVA and the probability of making a single Type I error in each family of tests remains greater than 0.05, such an approach is accepted by most researchers (Toothacker 1993).

Lipid class and protein content per egg and per nanolitre of egg volume were compared between species using nested ANOVA with species as a fixed factor and female as a nested factor. When a significant effect of species was detected, a conservative Tukey–Kramer test was employed for pairwise comparisons. 1-way ANOVA and a Tukey–Kramer test were used to examine between-species differences in egg lipid to protein ratios.

TG and protein utilisation in development to a feeding bipinnaria larva (Patiriella regularis, M. mortenseni) or nonfeeding brachiolaria larva (M. calcar) was examined using repeated measures ANOVA (with female as a random, between-subjects factor and cultures as the subjects). As eggs from three groups of females were used for rearing Parvulastra exigua, “group” replaced female as a random factor for this species. When a significant effect of time was detected, a single planned comparison was used to compare TG or protein content of unfertilised eggs with that of a fully developed bipinnaria (P. regularis, M. mortenseni) or brachiolaria larva (P. exigua, M. calcar). In addition, TG utilisation to the bipinnaria stage was calculated for each P. regularis and M. mortenseni culture and compared between species using nested ANOVA (with species as a fixed factor and female as a nested factor).

Results

Lipid classes present in eggs

Egg diameters, calculated egg volumes and lipid class data for the six asterinids are given in Table 1. Total lipid per egg differed immensely with egg size and ranged from 122 ng for the planktotroph Patiriella regularis to 8,980 ng for the lecithotroph M. gunnii (Table 1). For all species, a small peak was detected to the left of the KET internal standard on the first chromatogram (Fig. 1a). This peak co-occurred with the peak produced by spiking samples with a ME standard (methyl dodecanoate) and was consequently identified as ME (Fig. 1b). No WE was detected in the eggs of any species. If present, WE would have appeared to the left of the ME peak (Parrish 1999, Fig. 1).
Table 1

Patiriella regularis, Meridiastra spp. and Parvulastra exigua. Developmental mode, egg buoyancy, egg size (mean ± SE), and mean lipid and protein content per egg (ng egg1) and per nanolitre of egg volume (ng nl1)

 

P. regularis

M. mortenseni

M. oriens

M. calcar

M. gunnii

P. exigua

Development

Pt

Pt

PL

PL

PL

BL

Buoyancy

Negative

Negative

Neutral-Negative

Neutral-Negative

Positive

Negative

Egg diameter (μm)

165 ± 2.18

239 ± 4.19

399 ± 5.14

415 ± 3.18

431 ± 3.85

384 ± 2.69

Egg volume (nl)

2.4 ± 0.09

7.2 ± 0.47

33.5 ± 1.30

37.4 ± 0.84

42.0 ± 1.13

29.8 ± 0.62

Lipid classes (ng egg−1)

Energetic

  AH

1.3 (1.0%)

1.7 (0.6%)

85.5 (1.4%)

38.8 (0.6%)

11.8 (0.1%)

52.1 (1.3%)

  ME

1.7 (1.4%)

4.3 (1.4%)

87.7 (1.4%)

97.4 (1.5%)

191.5 (2.1%)

66.8 (1.7%)

  TG

41.3 (34.0%)

113.9 (36.3%)

4703.4 (76.9%)

5288.7 (80.1%)

7402.6 (82.4%)

2189.8 (56.0%)

Structural

ST

7.7 (6.3%)

13.7 (4.3%)

129.8 (2.1%)

144.7 (2.2%)

88.0 (1.0%)

146.5 (3.8%)

AMPL

8.1 (6.6%)

13.1 (4.2%)

318.1 (5.2%)

335.8 (5.1%)

198.7 (2.2%)

266.6 (6.8%)

PL

61.5 (50.7%)

167.1 (53.2%)

792.4 (13.0%)

693.2 (10.5%)

1087.9 (12.1%)

1187.2 (30.4%)

Total Lipid (ng egg−1)

121.5

313.8

6116.9

6598.7

8980.4

3909.0

Total Lipid (ng nl−1)

51.3

44.0

131.8

183.2

175.9

213.3

Protein (ng egg−1)

317.2

889.1

2359.4

3102.5

3018.2

4900.3

Protein (ng nl−1)

133.7

125.7

71.3

82.9

72.0

164.5

Data presented are calculated from means for five individual females of each species. Bracketed information for lipid class data gives percentage contribution to total egg lipid. Developmental mode abbreviations are: Pt planktotrophy, PL planktonic lecithotrophy, BL benthic lecithotrophy. Lipid class abbreviations are: AH aliphatic hydrocarbon, ME methyl ester, TG triglyceride, ST cholesterol, AMPL acetone-mobile polar lipid, PL phospholipid

https://static-content.springer.com/image/art%3A10.1007%2Fs00227-007-0809-7/MediaObjects/227_2007_809_Fig1_HTML.gif
Fig. 1

Meridiastra mortenseni and Meridiastra gunnii. Example TLC/FID chromatograms of egg lipid extracts analysed using the Parrish (1999) triple development. a For eggs of the planktotroph M. mortenseni, the small lipid peak to the left of the ketone standard on the first chromatogram was identified as ME by comparison with standards. b Spiking lipid extracted from M. mortenseni eggs with an ME standard (methyl dodecanoate) enlarged the ME peak and thus confirmed its identity. c Composite chromatograms (derived from the three chromatograms obtained) indicate that TG is the dominant energetic lipid in eggs of the planktotroph M. mortenseni. d TG is the dominant lipid in eggs of the lecithotroph M. gunnii. Lipid class abbreviations are: AH aliphatic hydrocarbon, KET ketone internal standard, TG triglyceride, ST cholesterol, AMPL acetone-mobile polar lipid, PL phospholipid, NLM non-lipid material

The same six lipid classes were detected in the eggs of all six asterinid species (Table 1, Fig. 1c, d). These consisted of three energetic lipids (AH, ME and TG) and three structural lipid groups (ST, AMPL and PL). Energetic lipids were dominated by TG in all species. For the two species with planktotrophic development, TG as a percentage of total egg lipid was similar at 34.0% in Patiriella regularis and 36.3% in M. mortenseni (Table 1). For the three species with pelagic lecithotrophic development, percentage egg TG was relatively constant and ranged from 76.9% in M. oriens to 82.4% for M. gunnii (Table 1). TG comprised 56.0% of total egg lipid for the benthic lecithotroph Parvulastra exigua (Table 1). Structural lipids were dominated by PL in all species and this lipid class ranged from 10.5% of total egg lipid in M. calcar to 53.3% in M. mortenseni (Table 1).

Maternal investment in structural and energetic lipid differed significantly between species (nested ANOVAs, F(5,24) = 123.5 and 1522.3, both P < 0.001; Fig. 2a, b) and between females within species (F(24,60) = 5.77 and 2.85, both P < 0.001). The large eggs of the four lecithotrophs contained more structural and energetic lipids than those of the two planktotrophs (Tukey–Kramer test; Fig. 2a, b). Differences in total energetic lipid were primarily due to oogenic deposition of TG which differed between species (nested ANOVA, F(5,24) = 1432.5, P < 0.001; Fig. 2c) and between females within species (F(24,60) = 2.89, P < 0.001). We focus on this lipid class for the energetic comparisons that follow. For the planktotrophs, TG per egg was greater in M. mortenseni (113 ng) than P. regularis (41 ng) and thus reflected egg size (Tukey–Kramer test; Table 1, Fig. 2c). Similarly, egg TG increased with egg diameter for the four lecithotrophic species studied (Table 1, Fig. 2c).
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-007-0809-7/MediaObjects/227_2007_809_Fig2_HTML.gif
Fig. 2

Patiriella regularis, Meridiastra spp. and Parvulastra exigua. Egg lipid content (ng egg−1) for: a total structural lipid: sum of ST, AMPL and PL; b total energetic lipid:sum of AH, ME and TG; c TG only. Y-axes are logarithmic to allow better visual comparison of developmental modes. Data are mean + SE and each bar represents an individual female (n = 3 per female). Species that do not share a superscript letter differ significantly (Tukey–Kramer test). Lipid class abbreviations are as defined in the text. Shading of columns represents developmental mode: white bar planktotrophic, grey bar planktonic lecithotrophic, dotted bar benthic lecithotrophic

The effect of egg size on lipid class differences between species was removed by standardising to egg volume. This is required to distinguish between oogenic deposition processes that scale isometrically with egg volume and those that have been modified in the evolution of lecithotrophic development (Villinski et al. 2002). When egg TG was standardised to egg volume, differences remained between species (nested ANOVA, F(5,24) = 309.9, P < 0.001) and between females within species (F(24,60) = 3.2, P < 0.001), and importantly species were grouped by developmental mode (Tukey–Kramer test; Fig. 3a). Eggs of the planktotrophs Patiriella regularis and M. mortenseni did not differ in TG density (mean = 17.4 and 16.1 ng TG nl−1, respectively) but had a significantly lower TG density than eggs of Parvulastra exigua (mean = 73.9 ng TG nl−1; Fig. 3a). TG density in the eggs of the three lecithotrophic Meridiastra species with pelagic larvae (range of means: 140.9–175.7 ng TG nl−1) was significantly greater than in P. exigua which has benthic larvae (Fig. 3a). These data contrast with those for volume-standardised egg content of total structural lipid (Fig. 3b). Although differences exist between species (nested ANOVA, F(5,24) = 4.56, P < 0.007) and between females within species (F(24,60) = 5.42, P < 0.001), post-hoc multiple comparisons yielded complicated results with no unambiguous trends for structural lipid with respect to developmental mode (Fig. 3b).
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-007-0809-7/MediaObjects/227_2007_809_Fig3_HTML.gif
Fig. 3

Egg lipid content standardised to egg volume (ng nl−1) for: a TG, b total structural lipid: sum of ST, AMPL and PL. Results of pairwise comparisons are complicated for structural lipid and are not shown, all other details are as for Fig. 2

Total egg protein

Total protein per egg ranged from 317 ng for Patiriella regularis to 4,901 ng for Parvulastra exigua (Table 1) and differed significantly between species (nested ANOVA, F(5,24) = 306.4, P < 0.001; Fig. 4a) and between females within species (F(24,60) = 2.23, P < 0.007). Protein per egg was significantly greater in M. mortenseni (889 ng) than P. regularis (317 ng), reflecting the difference in egg size (Tukey–Kramer test; Table 1, Fig. 4a). The eggs of the benthic lecithotroph P. exigua contained significantly more protein than those of Meridiastra species with pelagic nonfeeding larvae (Tukey–Kramer test; Fig. 4a), despite having the smallest eggs of these species (Table 1). When total egg protein is standardised to egg volume differences remain between species (nested ANOVA, F(5,24) = 27.8, P < 0.001) and between females within species (F(24,60) = 3.1, P < 0.001; Fig. 4b). Protein density in the eggs of the planktotrophs P. regularis and M. mortenseni did not differ (mean = 133.7 and 125.7 ng protein nl−1, respectively) and was not significantly different from that in P. exigua eggs (mean = 164.5 ng protein nl−1; Fig. 4b). However, density of egg proteins in these three species was significantly greater than for the three Meridiastra lecithotrophs (range of means: 71.3–82.9 ng protein nl−1; Fig. 4b).
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-007-0809-7/MediaObjects/227_2007_809_Fig4_HTML.gif
Fig. 4

a Total egg protein (ng egg−1), b total egg protein standardised to egg volume (ng nl−1). All other details are as for Fig. 2

Egg lipid:protein ratio

In order to investigate the potential influence of lipid content on egg buoyancy (Table 1), a ratio of lipid to protein was calculated for eggs of each female studied (n = 5 per species). Egg lipid:protein ratio differed significantly between species (one-way ANOVA, F(5,24) = 101.1, P < 0.001; Fig. 5). This ratio is significantly higher for the negatively buoyant, benthic eggs of P. exigua (mean = 0.80) than the small, negatively buoyant eggs of the planktotrophs Patiriella regularis and M. mortenseni (mean = 0.38 and 0.37, respectively; Fig. 5). The three Meridiastra lecithotrophs produced eggs with the greatest lipid:protein ratios (Fig. 5). This ratio was higher (although not significantly so) for the buoyant eggs of M. gunnii (mean = 2.98) than the neutral-negatively buoyant eggs of M. oriens and M. calcar (mean = 2.62 and 2.15, respectively; Fig. 5).
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-007-0809-7/MediaObjects/227_2007_809_Fig5_HTML.gif
Fig. 5

Egg lipid:protein ratio (mean + SE, n = 5 females per species). Shading of columns represents developmental mode: white bar planktotrophic, grey bar planktonic lecithotrophic, dotted bar benthic lecithotrophic

Utilisation of maternal provisions during larval building

Larvae of Patiriella regularis and M. mortenseni developed at the same rate. Feeding bipinnaria of P. regularis at 3 days are smaller than M. mortenseni larvae (mean length ± SE = 623 μm ± 7.7 and 503 μm ± 8.0, respectively; n = 10 per species). Both species used a substantial portion of egg TG to develop to this stage (RMANOVA, Table 2, Fig. 6a, b). Egg TG stores had significantly decreased from 41.5 to 19.4 ng in Patiriella regularis (t test, t6 = 7.38, P < 0.001) and from 111.1 to 72.9 ng in M. mortenseni (t6 = 7.01, P < 0.001) during development of the bipinnaria. Production of smaller P. regularis larvae required a lower amount of TG (22.1 versus 38.2 ng, nested ANOVA, F(1,4) = 11.0, P < 0.05; Fig. 6a, b) but this represented a greater percentage use of initial egg TG (53 versus 34%). Total soluble protein per individual did not change significantly over time for either P. regularis or M. mortenseni larvae (Table 2; Fig. 6c, d).
Table 2

Results of repeated measures ANOVA investigating change in TG and protein content (ng indiv1) from egg to larval formation in Patiriella regularis, Meridiastra mortenseni, Meridiastra calcar and Parvulastra exigua

Species

Source of variation

df

TG

Protein

F

P

F

P

P.regularis

Female

2,6

5.18

NS

0.62

NS

Time

3,18

19.16

<0.01

1.88

NS

Female × time

6,18

3.87

<0.013

0.64

NS

M. mortenseni

Female

2,6

25.00

<0.01

6.26

NS

Time

3,18

24.91

<0.001

5.14

NS

Female × time

6,18

2.21

NS

0.44

NS

M. calcar

Female

2,6

1.31

NS

0.61

NS

Time

3,18

0.23

NS

9.96

<0.01

Female × time

6,18

1.00

NS

0.11

NS

P. exigua

Group

2,6

0.80

NS

0.28

NS

Time

3,18

2.69

NS

9.92

<0.01

Group × time

6,18

1.44

NS

0.52

NS

The analyses included one between-subjects random factor (female or group), with cultures as the subjects and time as the repeated measure. Type I error rates were transformed to 0.013 for analyses of both variables. NS not significant, significant results highlighted in bold

https://static-content.springer.com/image/art%3A10.1007%2Fs00227-007-0809-7/MediaObjects/227_2007_809_Fig6_HTML.gif
Fig. 6

Patiriella regularis and Meridiastra mortenseni. a, b TG and c, d protein content (ng indiv−1) from egg to development of the feeding bipinnaria. Data are mean ± SE for each female (n = 3 per female) and each symbol denotes offspring from one female. Symbols are partially offset at each time point for clarity and obscure error bars in some cases. When a significant effect of time was detected (Table 2), a plotted line illustrates the temporal change in the mean value. Significant results from a planned comparison between unfertilised eggs and fully-formed larvae are shown: ** P < 0.01, *** P < 0.001

Over development to the feeding bipinnaria stage, TG per individual differed significantly between offspring of different females for Patiriella regularis and M. mortenseni (RMANOVA, Table 2, Fig. 6a, b). For P. regularis, the main effect of female is difficult to interpret because the female × time interaction term is also significant. While egg TG content differed between females, these differences disappeared by the bipinnaria stage. For M. mortenseni however, between-female differences in TG content were still evident in bipinnaria stage at 3 days (Fig. 6b). For example, mean TG content for eggs of one female was initially 29 ng lower than for a second (93 versus 122 ng egg−1) and remained 34 ng lower at the bipinnaria stage (52 versus 86 ng indiv−1; Fig. 6b).

For the pelagic and benthic lecithotrophs M. calcar and Parvulastra exigua, there was no evidence of TG being used to support development to the brachiolaria larval stage (Table 2, Fig. 7a, b). Egg proteins were depleted during larval formation in these species (Table 2, Fig. 7c, d). Egg proteins decreased by 9.7% to the 4 day-old brachiolaria in M. calcar (mean loss = 278 ng; t6 = 3.63, P < 0.01) and by 10.8% to the 5 day-old brachiolaria in P. exigua (mean loss = 521 ng; t6 = 3.82, P < 0.01).
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-007-0809-7/MediaObjects/227_2007_809_Fig7_HTML.gif
Fig. 7

Meridiastra calcar and Parvulastra exigua. a, b TG and c, d protein content (ng indiv−1) from egg to development of the nonfeeding bipinnaria. Each symbol denotes offspring of one female (M. calcar) or one “group” of females (P. exigua). All other details are as for Fig. 6

Discussion

Energetic component of asterinid egg lipid is triglyceride

Energetic egg lipids were dominated by TG in the planktotrophic asterinids, Patiriella regularis and M. mortenseni, as in other asteroids (Oudejans and van der Sluis 1979; Broertjes et al. 1985), echinoids (e.g. Deguchi et al. 1979; Yasumasu et al. 1984; Yokota et al. 1993; Sewell 2005; Meyer et al. 2007) and ophiuroids (Falkner et al. 2006) with feeding larvae. The four lecithotrophic species examined produced large eggs also dominated by TG while WE was not detected. Similarly, WE is present at no more than trace levels in ovaries of lecithotrophic asteroids from cold waters (Falk-Petersen and Sargent 1982) and TG is the dominant lipid present in eggs of the lecithotrophic ophiuroid Ophionereis schayeri (Falkner et al. 2006).

Our results contrast with one previous finding that the evolution of lecithotrophic development in the Asterinidae has involved increased oogenic deposition of WE (Villinski et al. 2002). The composition of echinoderm eggs and gonads is known to be affected by adult diet (George et al. 1991; George 1996; Watts et al. 1998) so feeding differences between sampling sites could potentially explain this discrepancy. However, the current data indicate the importance of maternal TG provisioning in asterinid species from warm and cold temperate latitudes in Australia and New Zealand. Hence the consistency of egg lipid profiles reported here is unlikely to be the product of habitat-specific adult diets. We therefore conclude that, regardless of developmental mode, TG is the primary maternally derived lipid fuel available for asterinid development.

Oogenic modification and developmental mode

Maternal provisioning in the asterinids studied was directly related to their developmental mode. Even after correcting for egg volume, egg TG reserves of the lecithotrophic species dwarfed those found in their planktotrophic relatives (Fig. 3a). There was no general increase in the density of egg structural lipids associated with the evolution of lecithotrophy. Evolution of a large lipid-rich egg appears to have been accomplished by hypertrophic elaboration of the ancestral-type, oogenic programme of TG deposition. Egg protein density was greater for species with pelagic, planktotrophic development than those with pelagic, lecithotrophic larvae. This is probably explained by the greater need of planktotrophic species to provision eggs with structural materials (proteins) suitable for building a functional feeding larva (Jaeckle 1995).

The asterinids with planktonic larvae spawned eggs with a range of buoyancies reflected in their lipid:protein ratios. The negatively buoyant eggs of the planktotrophs Patiriella regularis and M. mortenseni exhibited the lowest ratios (0.38 and 0.37, respectively) and these were within the range reported for other shallow-water, planktotrophic echinoderms (0.21–0.78, George et al. 1997). The Meridiastra species with planktonic, lecithotrophic larvae all produced eggs with lipid:protein ratios greater than 2.0 which reflects their greater egg buoyancy. This ratio was large for M. gunnii (2.98) so positive egg buoyancy in this species may be a by-product of selection for increased maternal investment in energetic lipid. Alternatively, this could reflect oogenic modification in response to selection favouring positively buoyant eggs if fertilisation can be maximised at the two-dimensional water surface (Harrison and Wallace 1990; Villinski et al. 2002).

The benthic lecithotroph Parvulastra exigua produced eggs with a low lipid:protein ratio (0.80) close to that reported for some planktotrophic echinoderms (George et al. 1997). Enhanced protein deposition in the eggs of this species may serve as a “weight-belt” to assist in maintaining a benthic location (Byrne 1995). Comparative histochemistry of the eggs of egg-laying and brooding asteroids similarly indicate a dominance of yolk protein (Chia 1968; Byrne et al. 1999a; Byrne and Cerra 2000). Interestingly, P. exigua eggs contained substantial amounts of phospholipid (mean = 1187 ng egg−1; Table 1), which is likely due to the abundance of yolk protein granules in the egg cytoplasm and the associated organelle membranes (Byrne et al. 1999a). These data indicate that for the lecithotrophic asterinids maternal provisioning has been influenced by selection for larval developmental habitat.

Maternal provisions fuel larval building

Egg TG reserves were depleted during development to the bipinnaria stage in Patiriella regularis and M. mortenseni with 53 and 34% being utilized, respectively. As is the case for planktotrophic echinoids (e.g. Yasumasu et al. 1984; Sewell 2005; Meyer et al. 2007), TG is an important lipid fuel used to support construction of a functional feeding larva in these asterinids. Maternal investment per egg is lower in P. regularis than M. mortenseni and the latter species uses its additional energetic reserves to produce a larger feeding larva. Similarly, planktotrophic echinoids with larger eggs produce larger early larvae with longer arms (McEdward 1986). Yolk proteins were not used as an energy source during larval construction in either P. regularis or M. mortenseni, which agrees with results for other planktotrophic echinoderms (e.g. Cellario and George 1990; Podolsky et al. 1994).

Neither M. calcar nor Parvulastra exigua utilised detectable amounts of TG in forming a nonfeeding brachiolaria larvae. Given that egg TG stores are not mobilised for larval building and that lipid droplets are visible throughout the development of these species (Byrne et al. 2003), the majority of maternally derived TG is presumably reserved for juvenile growth. This conclusion is supported by previous energetic studies of echinoderm ontogeny. For example, the facultatively planktotrophic echinoid Clypeaster rosaceus spawns far smaller eggs than the lecithotrophic asterinids investigated here, yet experimentally obtained quarter-size embryos of this species contain sufficient energy to support development to metamorphosis in the absence of food (Allen et al. 2006). Similarly, reduced-lipid embryos of the lecithotrophic sea urchin Heliocidaris erythrogramma metamorphose successfully and total lipid per individual decreases only after metamorphosis (Emlet and Hoegh-Guldberg 1997; Villinski et al. 2002). The egg TG reserves of lecithotrophic asterinids are likely to be particularly important during the perimetamorphic period (i.e. the time between metamorphosis and the opening of the larval mouth) (Gosselin and Jangoux 1998).

The construction of a nonfeeding brachiolaria larva in M. calcar and Parvulastra exigua may be partially fuelled by degradation and oxidation of yolk proteins. Subtracting the protein content of 3 day-old larvae from that of the egg indicates that rates of protein depletion are similar for these species (70 and 104 ng larva−1 day−1, respectively). Egg proteins are apparently used as an energy source for the nonfeeding development of some marine invertebrates (e.g. abalone: Moran and Manahan 2003) but few studies have considered protein utilisation in larvae of lecithotrophic echinoderms. Shilling and Manahan (1994) reared nonfeeding larvae of the Antarctic sea star Acodontaster hodgsoni at −1.2°C and reported that total protein decreased at a rate of 26 ng larva−1 day−1. Rate of protein depletion during larval construction appears to be greater in the lecithotrophic species examined here. Interestingly, total protein per larva is reported to increase over the development of the lecithotrophic sea star Mediaster aequalis, which may be due to uptake of dissolved organic material (Bryan 2004). Further studies are required to clarify whether protein is used as a significant fuel for development of asterinids with lecithotrophic larvae.

In planktotrophic echinoderm species, lipid reserves are accumulated by feeding larvae prior to metamorphosis (Cellario and George 1990; George et al. 1997; Reitzel et al. 2004). Although FFA content increases in developing echinoplutei (Sewell 2005; Meyer et al. 2007), to our knowledge no study has investigated the nature of lipid stores present in feeding echinoderm larvae that are competent to settle. The evolution of lecithotrophic development in the Asterinidae has involved a shift in the lifestage responsible for provisioning metamorphs with lipid fuels (i.e. the responsibility has shifted from a feeding larva to the female parent). We hypothesise that, despite considerable developmental evolution, juveniles of both planktotrophic and lecithotrophic asterinids are provided with the same lipid fuels. This hypothesis can be readily tested by rearing planktotrophic asterinid larvae to settlement and determining if they accumulate the same energetic lipid as that invested in the eggs of related lecithotrophic species (i.e. TG).

Implications for evolution of maternal investment in echinoderms

Regardless of developmental mode, maternal investment varied significantly between females within each asterinid species. For example, total lipid per egg ranged from 231 to 377 ng for the planktotrophic M. mortenseni and from 7,395 to 9,954 ng for the lecithotrophic M. gunnii. Clearly, all eggs of a given species are not created equal and egg quality is heavily affected by parentage as in other echinoderms (e.g. McEdward and Carson 1987; George 1994; Marshall and Bolton 2007) and other marine invertebrates (Phillips 2007). Between-female differences in investment per egg persisted after correcting for egg volume (Figs. 3, 4b) and therefore cannot be explained simply by differences in egg size. Rather, the packaging of lipid and protein into oocytes appears to be a variable trait as shown for mussels and barnacles (Phillips 2006).

Variation in maternal provisioning among females may be due solely to maternal effects that were not considered in this study such as maternal size, age or condition (Bernardo 1996). Alternatively, if there is a genetic component to the observed variation in egg quality this trait should be responsive to natural selection. We do not know if this is the case, although egg size in other marine invertebrates is highly heritable (Levin et al. 1991; Miles et al. 2007). Female-specific differences in TG per egg were still observable at the bipinnaria stage for M. mortenseni indicating that egg quality has flow-on effects for the nutrition of feeding larvae. If variation in maternal investment is heritable, selection for better larval nourishment may have favoured increases in maternal provisioning amongst planktotrophic asterinids and have contributed to the evolution of nonfeeding development in this group (Wray 1996).

Quantitative lipid class data obtained using the triple-development of Parrish (1999) and Iatroscan TLC/FID system are now available for the eggs of eight temperate echinoderms with planktonic larva (four planktotrophic and four lecithotrophic species). For the four planktotrophic species, energetic lipid (as a percentage of total egg lipid) is similar (ca. 40%) for eggs ranging in diameter from 87 μm for the echinoid Evechinus chloroticus to 239 μm for M. mortenseni (Fig. 8). For the four lecithotrophic species, percent energetic lipid is similar (ca. 80%) for eggs with diameters ranging from 248 μm for the ophiuroid Ophionereis shayeri to 431 μm for M. gunnii (Fig. 8). Although phylogeny is a potentially confounding factor, egg lipid profiles appear to be primarily a function of developmental mode whereas within each developmental type they are little affected by egg size. These data suggest that there may be two evolutionarily stable strategies of maternal lipid provisioning for temperate echinoderms with planktonic development.
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-007-0809-7/MediaObjects/227_2007_809_Fig8_HTML.gif
Fig. 8

Energetic egg lipid (as a percentage of total egg lipid) as a function of egg diameter for eight echinoderms with planktonic larvae. Data are included from the current study as well as for an echinoid (Sewell 2005) and two ophiuroids (Falkner et al. 2006). Symbols denote echinoderm class: open circle echinoid, inverted open triangle ophiuroid, open diamond asteroid. Mode of development is denoted by symbol shading: unshaded planktotrophic, shaded lecithotrophic

This dichotomy might be expected if evolution of egg lipid profiles is constrained: (1) in planktotrophic echinoderms by the structural requirements of feeding larvae (Wray 1996), (2) in lecithotrophic species by the energetic requirements of nonfeeding larvae and juveniles. It is also consistent with the hypothesis that lecithotrophic development has evolved from planktotrophy through a facultatively feeding larval stage. For species with levels of maternal investment near that sufficient for facultative planktotrophy, increased egg provisioning would have positive effects for juvenile nutrition similar to those demonstrated for lecithotrophic species (e.g. George 1994; Emlet and Hoegh-Guldberg 1997). Assuming egg quality variation is heritable, selection in this case would be expected to favour increased oogenic deposition of energetic lipid. This would most readily be accomplished by hypertrophic elaboration of an existing (ancestral-type) oogenic programme as seen in asterinids and echinometrid echinoids with lecithotrophic development (Byrne et al. 1999a, b). Under this scenario, intermediate levels of energetic lipid provisioning in echinoderm eggs will be rare because facultative feeding is an evolutionarily unstable developmental strategy.

Further data for free-spawning echinoderms are required to determine whether the disparity in egg lipid profiles between planktotrophic and lecithotrophic species is preserved (Fig. 8). It should be noted that energetic lipid proportions in the eggs of planktotrophic echinoderms may be greater for species from polar than temperate regions (Bosch et al. 1991; Podolsky et al. 1994). Lipid class data for the intermediate sized eggs of facultatively planktotrophic species will be especially valuable. Given the structural requirements of building a feeding larva, we predict that egg lipid profiles for these species will closely resemble those of obligately planktotrophic echinoderms.

Acknowledgments

Thanks to Inke Falkner for specimen and sample collection, Silver Bishop for help with lipid analyses, and Dr Kirsten Benkendorff for laboratory space. The Bosch Institute (University of Sydney) provided facilities for protein analyses. This manuscript benefited from the comments of three anonymous reviewers. This research was funded by a grant to MB from the Australian Research Council and complied with current laws regarding experimental science conducted in Australia and New Zealand.

Copyright information

© Springer-Verlag 2007