Establishing larval feeding regimens for the decorator crab Camposcia retusa : Effects of live prey types and density on larval survival and development

Decorator crab Camposcia retusa is a popular marine ornamental species; however, it has never been bred previously. To establish a feeding regime for C . retusa larvae, which include two zoeal and a megalopal stage, three experiments were conducted. In all experiments, ≥60% of unfed 1st zoeal (Z 1 ) larvae survived to the next stage, combined with the orange guts observed in larvae fed Artemia , suggesting Z 1 larvae are facultative lecithotrophic. Experiment 1 evaluated the suitability of ss-type rotifer Brachionus rotundiformis as prey. Z 1 larvae were

Such heavy reliance on wild-collected specimens combined with ill-managed fisheries in some developing counties as the major suppliers is likely to have negative impacts on the biodiversity of the fragile coral reef ecosystem (Olivotto et al., 2016). Marine ornamental aquaculture could alleviate the collection pressure on the reefs and, therefore, has been considered as an option that could balance the economic benefit while offers sustainability to the marine ornamental industry (King, 2019;Murray & Watson, 2014). Unfortunately, compared with finfish, relatively limited attention has so far been paid to the development of larval culture techniques for marine ornamental decapods (Calado et al., 2003).
Moreover, current research efforts on marine ornamental decapods captive breeding are biased towards ornamental shrimps.
As for ornamental crabs, the deficiency of information on larval culture is clear since only two species (i.e. Mithraculus forceps and Mithraculus sulptus) have been studied in detail for their larval rearing (Penha-Lopes et al., 2005;Rhyne et al., 2005).
Establishing an effective larval feeding regime is fundamental to the development of captive breeding techniques for any aquaculture species (Daly et al., 2009;Ruscoe et al., 2004). The type of live prey and their suitable feeding density are often two key considerations in the establishment of such a feeding regime (Calado et al., 2008;Tsuji et al., 2015). Rotifers Brachionus spp. and Artemia are two traditional live prey routinely used in brachyuran crab larval rearing Holme et al., 2009;Oliver et al., 2017;Ruscoe et al., 2004;Støttrup & McEvoy, 2003;Zeng & Li, 1999). However, rotifers and Artemia are both known to lack highly unsaturated fatty acids (HUFA), which are crucial to marine larval growth and development Støttrup & McEvoy, 2003;Suprayudi et al., 2004b). As the result, enrichment has been commonly practised to improve their nutritional values (Ferreira et al., 2008;Figueiredo et al., 2009;Støttrup & McEvoy, 2003). For example, Artemia enrichment was reported to significantly improve larval performance in various fish and crustacean species (Avella et al., 2007;Beder et al., 2018;Jinbo et al., 2013;Suprayudi et al., 2004aSuprayudi et al., , 2004b. For crabs, Artemia enrichment was reported to significantly improve larval survival of mud crab Scylla serrata (Suprayudi et al., 2004b) and accelerated larval development of horsehair crab Erimacrus isenbeckii (Jinbo et al., 2013). Therefore, enriching Artemia is generally recommended in the hatchery practice for marine crabs (Figueiredo et al., 2009;Suprayudi et al., 2004b). However, there are also reports suggested fatty acid enrichment of Artemia did not significantly improve larval performance in other crab species, including edible species of common spider crab Maja brachydactyla (Andrés et al., 2007), blue swimmer crab Portunus armatus (Basford et al., 2021), mud crab S. serrata (Williams et al., 1999) and marine ornamental crabs Mithraculus sculptus and Mithraculus forceps (Rhyne et al., 2005).
Compared with rotifers and Artemia, copepods are superior in nutritional value and digestibility (Ajiboye et al., 2011;Corner & O'Hara, 1986;Dhont et al., 2013;Støttrup & McEvoy, 2003); however, due to the difficulty and very low culture productivity of intensive culture of planktonic copepods (Alajmi & Zeng, 2014;Camus & Zeng, 2009), the utilization of copepods in the hatcheries is still limited (Drillet et al., 2011;van der Meeren et al., 2014). In fact, studies on copepod use in larval culture were mainly carried out for fish species, which generally obtained positive results (Barroso et al., 2013;Payne & Rippingale, 2000;Zeng et al., 2018). Similar studies on crustacean larvae are very few (Farhadian et al., 2009;Tang et al., 2020;Waiho et al., 2018). Among these limited studies, generally positive results were reported. For instance, the survival and growth of the black tiger shrimp Penaeus monodon postlarvae were significantly higher when co-fed cyclopoid copepod Apocyclops dengizicus with Artemia (Farhadian et al., 2009). Similarly, the megalopae of the Chinese mitten crab Eriocheir sinensis showed a significantly higher metamorphosis rate were fed wild-harvested frozen copepods with Centropages dorsispinatus as the dominant species (Tang et al., 2020).
The decorator crab Camposcia retusa is a popular marine ornamental crab species but has not been reported successfully bred previously (Calado et al., 2003). As the consequence, all individuals traded in the global aquarium market are currently sourced from the wild. The overall objective of the present study was to establish a reliable larval feeding regime for C. retusa. A series of experiments were conducted to investigate the suitability and appropriate feeding density of ss-type rotifer Brachionus rotundiformis and Artemia nauplii as prey for larval rearing of C. retusa, as well as the effects of copepod co-feeding with Artemia and Artemia enrichment on C. retusa larval performance.

| Broodstock maintenance
Decorator crab C. retusa broodstocks were purchased from Cairns Marine (Cairns Marine), a commercial marine ornamental collector and whole seller. After being air-freighted to James Cook University, Townsville, each pair of the broodstock was kept in 50-L tanks connected to a recirculation system (renewal rate at ca. 90 L/h). Half terracotta pots and rocks were placed in the tanks as shelters. All crabs were fed ad libitum with chopped prawns and mussels, as well as thawed blood worms twice daily. Throughout the experiments, the broodstock tanks were maintained at temperature 26-28°C, salinity 34-36, photoperiod L:D = 14 h:10 h, pH: 7.9-8.2, NH 4 + /NH 3 and NO 2 − < 0.25 ppm and NO 3 − < 5.0 ppm. Alajmi & Zeng, 2014). Briefly, the copepods were fed a 1:1 mixed algal diet of Isochrysis sp. and Chaetoceros muelleri and harvested daily during Experiment 2 to feed the larvae. When harvesting, the culture water was firstly drained through a 150µm mesh and then a 25µm mesh. The 150µm mesh retained copepodites and adults, while the 25µm mesh collected nauplii. Only the copepodites and adults were used for feeding crab larvae, while the nauplii collected were returned to the culture tanks. The copepod density estimation was similar to rotifers.

| General procedures
Under the aforementioned broodstock maintenance condition, egg incubation took 21-25 days. Once larval hatching was noticed in the morning, the positively phototactic larvae were first attracted to a strong light source and then gently scooped out with a beaker before being allocated randomly to replicates of different treatments of an experiment. According to our previous study, the larval development of C. retusa consists of two zoeal and a megalopal stage (Xu et al., 2019).
For all experiments, larvae were reared in 600-ml glass bea-

| Experiment 2:
Artemia nauplii feeding density and copepod co-feeding trial As the results of Experiment 1 showed that rotifer is not a suitable prey for C. retusa larvae, this experiment was designed to test the suitability of Artemia nauplii as prey for C. retusa larvae and its appropriate feeding density; in addition, effects of co-feeding Artemia with copepods P. crassirostris were evaluated. The newly hatched larvae of C. retusa were subjected to five treatments: the unfed control, Artemia nauplii provided at three densities of 5, 10 and 15 Artemia/ ml; and a co-feeding treatment of 5 Artemia + 5 copepods/ml. There were four replicates per treatment for this experiment.

| Experiment 3: Effects of Artemia enrichment trial
This experi-ment was to determine whether fatty acid enrichment of Artemia improved larval performance. In this experiment, five treatments were set up: unfed control; larvae fed either newly hatched Artemia nauplii or enriched metanauplii at 5 and 10 ind./ml respectively. Each treatment had four replicates.

| Data analysis
Data are expressed as mean ± standard error (SE). 'Larval surviving time (LST) of Z 1 ' and 'LST of Z 2 ' were calculated for all treatments of the rotifer feeding trial (Experiment 1) and the unfed controls of the other two experiments. LST of Z 1 and LST of Z 2 were defined as the average surviving time (days) from hatching to the death of all Z 1 or Z 2 larvae in a treatment respectively. LST of Z 1 (Z 2 ) was calculated using the formula: where N t is the number of dead Z 1 (Z 2 ) larvae recorded on day t of an experiment; and n represents the days that dead Z 1 (Z 2 ) larvae were found.
Larval surviving time (days) of Z 1 ( Since the results of the current study showed that the Z 1 larvae of C. retusa are facultative lecithotrophic, the survival of Z 1 larvae and mean surviving time of unfed larvae can be used as an indicator for larval quality. To compare larval quality between three batches used in different experiments in this study, the Z 1 survival and mean surviving time of unfed larvae from these experiments were analysed using one-way ANOVA, followed by Tukey's HSD test. A Kaplan-Meier test was run to compare the survival trend between the unfed control and different rotifer feeding density treatments in Experiment 1. Differences between survival curves were detected by a pairwise comparison using the log-rank test from the package 'survival' of the R language (Therneau, 2020).
Larval survival and average developmental duration of different treatments in Experiments 1 and 2 were analysed using one-way ANOVA followed by Tukey's HSD post hoc test. A two-way ANOVA was performed to analysis data of Experiment 3. The normality of the data and homogeneity of variance were tested prior to the ANOVA analysis. In addition, to compare differences of Z 1 survival between the unfed control and each feeding treatment in Experiment 3, a general linear model was applied. The percentage survival data were transformed using arcsine square root transformation to meet the assumptions of the analysis methods. For all experiments, unfed control was not included for statistical analysis from Z 2 onwards due to no larvae of the controls survived beyond Z 2 . A statistical probability of p < 0.05 was accepted as significant. All statistics were performed using R language (R 4.0.3; R Core Team, 2020).

| Facultative lecithotrophic of newly hatched larvae and larval batch quality comparison
Survival of newly hatched Zoea 1 (Z 1 ) larvae to the next stage (Zoea 2 or Z 2 ) in the unfed controls of all three experiments was high (≥60%), confirming lecithotrophic of the larvae. When larval survival of the unfed controls from the three experiments is compared, Experiment 2 (85.5 ± 2.9%) had significantly higher survival than the other two experiments (65.0 ± 5.0% and 60.0 ± 2.0% for Experiment 1 and 3 respectively; Table 1). In addition, although 'larval surviving time' (LST) of Z 1 larvae from the unfed controls of the three experiments was similar (3.2-3.5 days, Table 1, p = 0.771), LST of Z 2 larvae of the Experiment 2 control was the longest (4.6 ± 0.3 days) and significantly longer than that of Experiment 3 control (3.8 ± 0.1 days, p = 0.036). The mean Z 1 duration of the larvae successfully moulted to Z 2 in the three experiments ranged from 1.5 to 1.8 days, and the larvae of Experiment 2 again was the fastest in their development (1.5 ± 0.1 days) although no significant difference was detected among three batches of larvae (Table 1, p = 0.483).

| Experiment 1: Rotifer feeding trial
The daily survival of C. retusa larvae of different rotifer density treatments showed a similar trend, and no significant difference was detected among them (Figure 1 3.3 | Experiment 2: Artemia nauplii feeding density and copepod co-feeding trial Daily larval survival was similar across all treatments during the first 2 days; however, survival of the unfed control declined rapidly from day 3 onwards with a total mortality on day 7, while larval survival of all fed treatments was higher than 20% with the highest survival of 60.0% found the 10 Artemia/ml feeding treatment ( Figure 2). Indeed, between day 5 and 10, larvae survival of the 10 Artemia/ml treatment was consistently significantly higher than other treatments (p < 0.05).
However, from day 10 onwards, larval mortality of the10 Artemia/ml treatment also increased sharply, which eventually declined to a level only marginally higher than other treatments on day 12 when the experiment was terminated (p > 0.05; Figure 2 and Table 3).
When survival is evaluated as the percentage of larvae successful moulted to the next stages, the survival of Z 1 larvae to Z 2 stage was very high (>85.5%) across all treatments with the highest survival of 93.8 ± 2.4% found in the treatment that Artemia nauplii were given at 10 Artemia/ml, but no significant difference was detected between any treatments, including the unfed control (Table 3). Z 2 survival to the subsequent megalopal stage was also high for all treatments (>70%) except the unfed control, which suffered a total mortality. Among fed treatments, Z 2 survival of the 10 and 15 Artemia/ml feeding treatments was significantly higher than those of the 5 Artemia/ml and 5 Artemia + 5 copepod/ml treatments (p < 0.05; Table 3).
In terms of overall zoeal survival, it was also high (>60%) across all fed treatments, and the highest survival achieved was again from the 10 Artemia/ml feeding treatment (91.3 ± 2.7%), which was significantly higher than all other treatments (Table 3). However, despite high zoeal survival, survival of megalopae was poor across all treatments (<20%), resulting in low overall larval survival to the first crab stage (C 1 ). The best larval survival from Z 1 to C 1 was from the 10 Artemia/ml treatment (11.3 ± 3.8%), which was not significantly different from that of 15 Artemia/ml (7.5 ± 4.8%) and 5 Artemia + 5 copepods/ml (3.8 ± 2.4%) co-feeding treatment, but significantly higher than 5 Artemia/ml treatment (1.3 ± 1.3%).
Throughout the experiment, no significant difference was found in larval survival between the 5 Artemia/ml and 5 Artemia + 5 copepods/ml treatments (Table 3).
In terms of larval development, Z 1 duration was not significantly different among all treatments (Table 3). However, for the overall zoeal development (Z 1 to M), the two higher Artemia feeding density treatments (10 and 15 Artemia/ml) were significantly shorter when compared to 5 Artemia/ml and 5 Artemia + 5 copepod/ml treatments (p < 0.05; Table 3). For the overall larval development duration (Z 1 to C 1 ), due to only a single larva survival to C 1 in the 5 Artemia/ml treatment, the treatment was excluded from the statistical analysis.
Of the remaining three fed treatments, the overall larval duration was again significantly shorter in the 10 and 15 Artemia/ml treatments when compared to that of 5 Artemia + 5 copepod/ml treatment (Table 3).

| Experiment 3: Effects of Artemia enrichment trial
For the first 3 days, daily larval survival showed a very similar pattern among all treatments; however, similar to the previous experiments, survival in the unfed control plummeted and a 100% mortality occurred on day 6 ( Figure 3). Of the fed treatments, between day 4 and 6, larval survival dropped sharply in all treatments except the  In terms of larval survival based on instar, Z 1 survival of the unfed control was the lowest but was not significantly different from all Artemia feeding treatments ( Figure 4A). Similar to the result in Experiment 2, among fed treatments, the best survival to Z 2 was found in the treatments in which larvae were fed Artemia at 10 ind./ ml, either in the form of unenriched nauplii (76.3 ± 2.4%) or enriched metanauplii (72.5 ± 4.8%; Figure 5A), although the difference was not significant compared with lower Artemia feeding density treatments at 5 ind./ml (F 1,12 = 0.179; p = 0.680). However, from Z 2 onward, the differences in larval survival became significant: highly significant effects of both Artemia density and enrichment on Z 2 survival were detected (p < 0.01) although no significant interaction of the two factors was found ( Note: Data are presented as mean ± standard error. Z 1 : first stage zoea; Z 2 : second stage zoea; M: megalopa; C 1 : first stage juvenile. Means with different superscript letters within a same row indicate significant differences (one-way ANOVA; p < 0.05).
*Statistical analysis did not include 5 Artemia/ml treatment due to only one larva survived to C 1 of the treatment.

F I G U R E 3
Daily larval survival of decorator crab Camposcia retusa fed Artemia at two densities (5 and 10 ind./ml) in the forms of unenriched nauplii and enriched metanauplii respectively. For clarity of the figure, the standard error bars are omitted

F I G U R E 4 Survival (A) and development duration (B) of the first
stage zoeal larvae of Camposcia retusa from the unfed control and the treatments in which larvae were fed Artemia at two densities (5 and 10 ind./ml) in the forms of unenriched nauplii and enriched metanauplii respectively. Data are presented as general linear model: A line within the box marks the median value; a cross within the box marks the mean value, the box spans the 25% and 75% while the bar represents the minimum and maximum value of a treatment. No significant difference was detected between unfed control and feeding treatments the overall zoeal survival (i.e. Z 1 to megalopae; 50.0 ± 6.1%), and both survivals were significantly higher than all other treatments ( Figure 5B,C). However, similar to Experiment 2, high mortality occurred during megalopal stage in all treatments, resulting in the best survival to C 1 at 7.5 ± 2.5% from larvae fed 10 ind./ml unenriched Artemia treatment; two-way ANOVA detected no significant difference in overall larval survival from Z 1 to C 1 related to either Artemia density, enrichment or their interaction (Table 4).
In terms of larval development, two-way ANOVA detected no significant effect of either Artemia density or enrichment, nor the interaction of the two factors on Z 1 and Z 2 development durations (p > 0.05; Table 4; Figures 5B and 6A,B). For overall larval development from Z 1 to C 1 ( Figure 6C), no statistics were performed due to only 1 datum was obtained in several treatments as the result of low survival.

| DISCUSS ION
The results of the three larval experiments showed that when not fed at all, newly hatched Z 1 larvae of C. retusa could still obtain a survival of between 60.0% and 83.8% to the Z 2 stage, confirming   (Anger, 2001;Zeng et al., 2020). Based on such definitions, C. retusa Z 1 larvae should be considered as facultative lecithotrophy because in both Experiment 2 and 3, it was noticed that in all Artemia feeding treatments, the guts of Z 1 larvae turned orange after a while.
Additionally, for both Experiments, Z 1 survival of the unfed control was always lower than any Artemia feeding treatments. Furthermore, Facultative lecithotrophy is generally considered as an adaptive mechanism evolved to cope with low or unpredictable food availability in the environment where larval development takes place (Anger, 2001;Tapella et al., 2012;Zeng et al., 2020). It is hence more commonly found in environments characterized with poor or unpredictable food availability, such as high-latitude marine waters or freshwater (Anger, 2001). The current finding that C. retusa Z 1 larvae are facultative lecithotrophy is therefore interesting, since C. retusa is considered a tropical species (WoRMS, 2021). Nevertheless, even environments that are normally considered as high productivity may also experience temporal or spatial variations in plankton availability, leading to a transitory or localized food limitation (Constable et al., 2003;Mackas et al., 1985). well-known phenomenon in aquaculture, which can significantly affect larval rearing outcomes (Calado & Leal, 2015;Mann et al., 1999).
The current results suggested that larvae used for Experiment 2 were likely of better quality than the other two batches used in Experiment 1 and 3, and the larvae used in Experiment 3 are probably of the lowest quality of the three. The difference in larval quality between batches is further confirmed by the major difference in larval survival when those larvae reared under a same feeding regime in two experiments are compared. For example, when C. retusa larvae fed Artemia nauplii at 10 ind./ml, survivorship to megalopa reached 91.3 ± 3.1% in Experiment 2, but it was only 50.0 ± 6.1% in Experiment 3 (Table 3 and Figure 4).
While as the result of facultative lecithotrophy, high percentages (>60%) of Z 1 larvae in the unfed controls of all three experiments successfully moulted to Z 2 , none of them survived to the next stage as megalopa. This result showed that from the Z 2 onwards, food availability, as well as their quality and quantity become crucial to the survival of the larvae. In Experiment 1, larvae were fed rotifers at a density ranging from 30 to 90 ind./ml; none of the Z 2 larvae in any of the treatments survived to megalopae, with total mortality occurred within 8 days regardless of rotifer feeding density. In addition, Z 1 survival of the fed treatments (56.7%-68.3%) and the unfed control (65.0%) was very similar (Table 2). Despite rotifers have been shown to be an appropriate diet for newly hatched larvae of a range of portunid and spider crabs (Baylon, 2009;Kogane et al., 2007;Nghia et al., 2007;Suprayudi et al., 2002;Zmora et al., 2005), our results suggest that ss-type rotifers are not a suitable diet for C. retusa larvae, which likely due to too small size of the rotifers. In this study, the super-small type or ss-type rotifer Brachionus rotundiformis was used, which has a size range of 90-160 µm; such a size range is likely F I G U R E 6 Effects of the Artemia enrichment and feeding density on the larval development of Camposcia retusa. (A) Zoea 1 (Z 1 ) duration; (B) Zoeal duration (Z 1 to megalopal stage); (C) Overall larval duration (Z 1 to 1st crab stage). *Only 1 datum obtained from each of these treatments too small for the C. retusa larvae to either capture or despite being consumed, energy budget was poor so that could not sustain successful development to the next stage. Similarly, in southern rock lobster, Jasus edwardsii, phyllosoma larvae showed prolonged development when live prey was too small (Ritar et al., 2003). Indeed, the relationship between prey size and larvae is a critical factor to consider in larval feeding, the disproportion of the size between food particles and larval feeding appendages may incapacitate the larvae to effectively capture the prey or despite larvae can capture and consumed the prey, they would eventually die due to unable to acquire a sufficient energy to sustain normal development (Guarizo et al., 2020;Ritar et al., 2003;Ruscoe et al., 2004;Sui et al., 2008).
Since copepods are known to have superior nutritional value than rotifers (Støttrup & McEvoy, 2003), a treatment using copepods co-fed with Artemia nauplii was tested in Experiment 2. Although copepods have been reported to be superior to rotifers and Artemia for fish hatchery (Barroso et al., 2013;Payne & Rippingale, 2000;Zeng et al., 2018), few previous research reported using copepods as sole prey for decapod larval culture, probably due to their very low culture productivity (i.e. typically <10 ind./ml for planktonic copepods vs >1000 ind./ml for ss-rotifers). On the contrary, co-feeding copepods with other traditional prey in decapod larval rearing have been trailed and reportedly enhanced survival and growth in several decapod species, including mud crab Scylla olivacea (Jantrarotai et al., 2004), Chinese mitten crab E. sinensis (Tang et al., 2020) and black tiger prawn Penaeus monodon (Farhadian et al., 2009). However, in the present study, co-feeding 5 copepods/ml with 5 Artemia/ml not only did not produce any clear beneficial effects, but both larval survival and development were significantly inferior when compared to those fed 10 Artemia/ml. In fact, larval survival and development of the copepod co-feeding treatment were similar to the treatment that Artemia nauplii were fed at 5 Artemia/ml (Table 3), suggesting that copepod P. crassirostris feeding hardly offered any additional benefits to C. retusa larvae. There were two possible explanations for the ineffectiveness of copepod co-feeding, firstly the length of copepodite and adult P. crassirostris ranging between 195 and 445 μm (Alajmi et al., 2015); thus, most of them are smaller than Artemia nauplii (400-500 μm). As mentioned previously, the disproportion of the size of the prey and the mouthpart of larvae may render it difficult for the larvae to capture the prey (Guarizo et al., 2020). Secondly, the fast and zigzag spurting swimming behaviour characterized by calanoid copepods may make it difficult for the larvae to successfully capture and consume them (Lumasag et al., 2007).
In contrast to rotifers and copepods, the results of Experiment 2 showed that in all treatments in which Artemia nauplii were fed to the larvae, larval survival to megalopae was higher than 60%. Thus, it proved that Artemia is an appropriate prey for zoeal larval rearing of C. retusa. In particular, larvae fed Artemia nauplii at 10 Artemia/ml have significantly better survival and development to the megalopae than those fed at a lower density of 5 Artemia/ml, suggesting Artemia feeding density is also important, which can be explained by the fact that decapod larvae are considered 'inactive predators' and their food consumption often relied on the chance of encounters with the prey (Jeffs & O'Rorke, 2020;Tsuji et al., 2015;Zeng & Li, 1999).
Indeed, previous studies have shown a positive relationship between ingestion rate and prey density until a saturation level is reached in various decapod larvae, including crabs E. sinensis (Sui et al., 2009), Hyas araneus (Anger & Dietrich, 1984), Ranina ranina (Minagawa & Murano, 1993) and shrimps Lysmata wurdemanni (Zhang et al., 1998), Penaeus kerathurus (Yúfera et al., 1984) and Penaeus monodon (Loya-Javellana, 1989). The importance of prey density is probably particularly relevant to early larvae who typically have weak swimming ability (Lumasag et al., 2007). Interestingly, in the same experiment, as Artemia density further increased to 15 ind./ml, larval survival decreased significantly. This probably can be explained by excess number of Artemia degraded water quality (Sui et al., 2009;Zhang et al., 1998).  also reported that an excessive high Artemia density could induce morphology abnormality in larvae of the last stage zoeal larvae of swimming crab Portunus trituberculatus, leading to mortalities during metamorphose. Therefore, based on the results of the current study, Artemia feeding density of 10 Artemia/ml appears to be most appropriate for C. retusa larvae.
Although larval survival to megalopae was relatively high in Experiment 2, mass mortality still occurred during the megalopal stage. In particular, for the 10 Artemia/ml treatment that had maintained a relatively high zoeal survival, mortality increased sharply during the period when megalopae metamorphosed to the first crabs (i.e. days 8-10; Figure 2). Since malnutrition could be a major cause of mass mortality of megalopae observed, Experiment 3 was designed and conducted to test whether Artemia enrichment may improve megalopal survival. Artemia enrichment has been reported to enhance larval survival and development of various decapod species (Dan, Oshiro, et al., 2016;Dan, Sui, et al., 2016;Suprayudi et al., 2004b). For example, in swimming crab P. trituberculatus, it was shown that larvae fed microalgae Nannochloropsis enriched Artemia achieved substantially higher megalopal survival (40.7%) than those fed unenriched Artemia (1.9%-3.5%; Dan, Oshiro, et al., 2016).
Likewise, mud crab S. serrata larvae fed unenriched Artemia were reported to exhibit reduced survival and prolonged development to C 1 stage (Suprayudi et al., 2004b). However, in the present study,  (Williams et al., 1999).
In the cases of marine ornamental crabs, Rhyne et al. (2005)  hatched Artemia nauplii are lecithotrophic; therefore, prior to their development to the next metanaupliar stage when filter-feeding starts, which also enable enrichment, the nutrition/energy content of the nauplii is continuously utilized for maintenance and, therefore, will reduce substantially when they moult to metanuaplii (Navarro et al., 1999). Indeed, for the GSL Artemia strain used in this study, it was reported that metanauplii had a 34% reduction in individual dry weight and a 37% reduction in energy content compared with newly hatched nauplii (Vanhaecke et al., 1983). Since commonly used oil emulsions are aimed at fatty acid enrichment, they often do not supplement with proteins or other nutrients. In fact, it has been reported that enriching GSL Artemia with an oil emulsion (DHA Selco, INVE) similar to the one used in this study led to a 19% reduction in proteins compared with newly hatched nauplii (Evjemo et al., 2001). Proteins are known as a key nutrient for crustaceans, especially during larval development when rapid tissue synthesis occurs (Anderson & De Silva, 2011;Anger, 2001). Hence, it is possible that in the present study, when Artemia metanauplii were enriched with S.presso, which is known to be rich in lipids but with a limited protein content, an unbalanced protein/lipid ratio may result, which led to inferior survival and development in those zoeal larvae fed the enriched Artemia.
Multiple factors have been reported to cause mass mortality of megalopae during larval rearing of various crab species, which include moult death syndrome, cannibalism, lack of appropriate settlement cues and substrates, and inappropriate physical culture environment (Beder et al., 2018;Dan, Sui, et al., 2016;Hamasaki et al., 2002bHamasaki et al., , 2007Jinbo et al., 2013).
For example, in this study, it was observed that up to 51% of dead megalopae missing limbs or body parts, suggesting cannibalism may play a major role in the megalopal mortality. Hence, steps taken to reduce cannibalism during megalopae rearing, such as reduced stocking density and providing shelters, may help improve megalopae survival. Additionally, the inappropriate cultural environment, such as the lack of suitable substrate and settlement cues, could also negatively affect megalopae metamorphosis (Forward et al., 2001), which has been confirmed by further studies (Xu, 2021).

| CON CLUS ION
Through a series of three experiments, this study investigated suitable larval feeding regime for captive breeding of decorator crab C. retusa, a popular marine ornamental crustacean. This is the first report on successful rearing of the species from newly hatched larvae to the first juvenile stage, and survival up to 91.3% to megalopal stage and 11.3% to the first juvenile stage, was achieved respectively. The results suggested that newly hatched C. retusa Z 1 larvae are facultative lecithotrophic, who are capable of developing to the next stage independent of food but can also perform exogenous feeding when food is available. It also reveals that unlike many crab species, sstype rotifer is not an appropriate diet for C. retusa zoeal larvae as it led to total mortality at Z 2 stage. In contrast, Artemia nauplii were found to be a good diet for the larvae while the feeding density of Artemia is also important. An Artemia feeding density of 10 ind./ml was identified as performing the best. Meanwhile, both co-feeding Artemia with copepods and Artemia enrichment were not found to improve larval survival and development, therefore, unnecessary.

ACK N OWLED G EM ENT
This work forms a part of the PhD thesis of the first author. Special thanks to Emeritus Professor Rhondda Jones for her help and advice on data analysis.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

AUTH O R CO NTR I B UTI O N S
Experiment design, conducting experiments, data collection and analysis, writing manuscript: Tian Xu. Supervision, experiment design, providing experimental materials, and revising manuscript: Chaoshu Zeng.

DATA AVA I L A B I L I T Y S TAT E M E N T
All relevant data are within the paper.

R E FE R E N C E S
Ajiboye, O., Yakubu, A., Adams, T., Olaji, E., & Nwogu, N. (2011). A review of the use of copepods in marine fish larviculture. Reviews in