Origin and maintenance of prawns

Adult M. amazonicum were obtained from the Aquaculture Center of the State University of São Paulo, Jaboticabal, Brazil (CAUNESP; for details of broodstock production and maintenance, see ^{Moraes-Valenti & Valenti, 2007}). The broodstock originated from estuarine tidal creeks near the city of Belém in the Amazon Delta in northern Brazil (01°14′30″S, 48°19′52″W), close to the type locality of this species (W. Valenti, pers. comm.). Male and ovigerous female shrimps were transported by air freight from Brazil to the Helgoland Marine Biological Laboratory (BAH), Germany, maintained in cooling boxes with aerated fresh water. The ovigerous females were subsequently transferred to recirculating aquaria provided with 30 L aerated freshwater (total ion concentration <0.2 mg L⁻¹) and gravel filters, kept at a temperature of 29 ± 1°C, a 12:12 h light:dark cycle, and fed with pieces of frozen marine isopods (Idotea spp.) and commercial aquarium feeds (Novo Tab, JBL; ^{Anger et al., 2009}). Sieves (0.3 mm mesh size) receiving the overflowing water from the aquaria were checked daily for the appearance of newly hatched larvae.

Larval rearing

Actively swimming newly hatched larvae from four females were transferred to aerated beakers with 1 L filtered water and subsequently reared under four experimental conditions with two salinities (5, 10) and two different types of salt (artificial vs. natural). Different salinities were obtained by mixing appropriate amounts of tap water (total ion concentration 0.2 mg L⁻¹) with natural seawater from the southern North Sea (salinity ca. 33) or with refined industrial sea salt (Tropic Marin salts; Wartenberg, Germany), respectively. Salinities were measured to the nearest 0.1 using a temperature-compensated electric probe (WTW Cond 330i, Weilheim, Germany). Conditions of temperature and light were the same as in the maintenance of adult shrimps. The cultures were checked every 24 h for molts or mortality when water and food (freshly hatched brine shrimp nauplii, Artemia franciscana; Great Salt Lake stock; ca. 10-15 nauplii mL⁻¹) were changed. A recent study revealed that a daily feeding interval and high food density (4-12 nauplii mL⁻¹) allow for optimal larval survival and growth (^{Maciel et al., 2014}). No food was given to the first zoeal stage because it is fully lecithotrophic (^{Anger & Hayd, 2009}).

Sampling, measurements of dry weight (W), elemental (C, N) and proximate biochemical composition (total lipid, protein)

The larval development of M. amazonicum shows individual variability in the number and duration of successive larval instars, normally comprising 8-10, occasionally up to >12 molts, before the first juvenile stage is reached ca. 20-22 days after hatching (^{Anger et al., 2009}). The change from late larval to early juvenile stages is in this species, typically for caridean shrimps, a gradual transition without metamorphosis (^{Anger, 2001}). The first juvenile was here defined as the first postembryonic stage where larval traits (e.g., natatory exopods on the pereiopods; planktonic swimming behavior) become functionally insignificant, while juvenile-adult traits clearly predominate (e.g., benthic crawling behavior, use of pereiopodal endopods as walking legs) (^{Anger, 2001}). In the present study, we quantified larval growth under different experimental conditions only for the most commonly observed developmental pathway, which consisted of nine larval stages followed by a juvenile. All larvae were therefore microscopically staged and sorted in daily intervals so that each rearing beaker (n = 4) contained homogeneous material with individuals that molted the same day to the same morphological stage; in each treatment, developmentally retarded larvae were removed from cultures and not used for analyses. The average time of molting to successive stages was slightly delayed in treatments with low salinity (5) and artificial salt in comparison to salinity 10 and natural sea salt, reaching in total about two days at the end of the experiment. This slight delay was not quantified in this study; likewise, no mortality rates were determined, as survival was generally high (>80%).

Each successive stage was sampled one day after molting; this implies that the time of sampling was determined by the time after each molt (stage C of the molting cycle; ^{Hayd et al., 2008}) and not by the absolute age from hatching. For each sample, parallel determinations of dry weight (W) and contents of carbon (C), nitrogen (N), total lipid, and protein were carried out with standard techniques (^{Urzúa & Anger, 2011}; ^{Urzúa et al., 2012}), considering these constituents as proxies of organic matter. Moreover, carapace length (CL) was microscopically measured in first-stage juveniles, indicating the increment in body size achieved during the period of total larval development. All measurements of W, C and N comprised five replicate samples with 2-5 individuals each, depending on individual W. For proximate biochemical analyses, four replicate determinations were carried out with 12-15 individuals each.

Statistical analyses

Statistical analyses were performed with standard methods (^{Sokal & Rohlf, 1995}; ^{Quinn & Keough, 2002}) using the software package Statistica 8 (StatSoft). Differences in W, C, N, lipid, and protein (response variables) in different developmental stages were tested using a 4-way ANOVA (nested design) with “Salinity” and “Type of salt” as independent factors, and “Female” and larval “Stage” as nested factors. Differences in body size (CL) of first-stage juveniles were tested using a two-way ANOVA with salinity and type of salt as factors. Where significant differences occurred, these were further analyzed with multiple pairwise comparison tests [Post-hoc test Student-Newman-Keuls (SNK tests)]. All tests were run on the 95% confidence level (P < 0.05). Normality and homogeneity of variances were tested with Kolmogorov-Smirnov and Levene tests, respectively.

Larval growth (biomass and body size per individual)

The first two zoeal stages of M. amazonicum are fully or facultative lecithotrophic, respectively (see ^{Anger & Hayd, 2009}). Compared with the Zoea-I stage (hatched in fresh water and sampled at the beginning of larval exposure to different experimental treatments), the Zoea II (1 day after molting) showed, therefore, no growth, regardless of salinity and type of salt (Table 1). Subsequent developmental instars, by contrast, showed in all measures of biomass per individual, and in all treatments, exponential patterns of increase (Table 1).

The rate of growth tended to be slightly lower in treatments with low salinity (5 vs. 10) and in those with artificial vs. natural salt. Statistically significant differences between treatments were more consistently found in dry weight (W) and amounts of elemental constituents per individual (C, N; Table 1) than in biochemical parameters (lipid, protein; Table 1). In the latter, effects of salinity were generally more conspicuous than those of the type of salt.

The tendencies observed in larval growth under different experimental conditions were clearly visible also in differential body size of first-stage juveniles (Table 2). Mean carapace length (CL) was significantly smaller after larval rearing at salinity 5 compared to 10; at both salinities, it was significantly smaller in treatments with artificial instead of natural salt. Among the conditions tested here, a combination of low salinity (5) with artificial salt (treatment 5A) was pessimum, while salinity 10 prepared by dilution of natural seawater (treatment 10N) represented the optimum among the rearing conditions tested here.

In spite of being statistically significant, the reduction of larval growth under suboptimal conditions was in general small. When natural sea salt was used, a reduction of salinity alone (treatment 5N vs. 10N) caused a reduction in early juvenile size by 11%. At salinity 10, the use of artificial salt reduced the CL by 8% (treatment 10A vs. 10N). The maximum reduction in body size (19%) was observed under pessimum compared to optimum rearing conditions (treatment 5A vs. 10N).

Changes in elemental and proximate biochemical composition

Compared to values of biomass per larva, those of relative elemental and biochemical composition (in % of W; C/N mass ratio) revealed less clear patterns of developmental change (Table 3). The percentage C content and the C/N ratio decreased initially (larval stages I-V), remained subsequently rather stable, and increased slightly in late larval stages and the first juvenile. The N content showed a weak tendency to increase during early development (larval stages I-V), remaining rather stable thereafter.

The percentage values (in % of W) of C and the C/N mass ratio tended to be lower in treatments with artificial vs. natural salt and at salinities 5 vs. 10. By contrast, the N content was mostly maximum in the treatment with artificial salt and salinity 5, without showing a consistent tendency in the other treatments.

The percentage values of lipid and protein showed no consistent response to different salinities or types of salt. The lipid content increased slightly, from initially about 14% to ca. 16-19% of W in larval zoea VI, remaining stable thereafter. Maximum values were observed at salinity 10, regardless of natural or artificial salt. A minimum lipid value (11% of W) was measured in early juveniles (Table 3). The percentage of protein increased from about 36% after hatching to >50% in later larval stages. A maximum protein level was found in first-stage juveniles.

Statistical analysis of developmental and treatment effects on larval growth and chemical composition of biomass

An overall comparison of the effects of the experimental factors “Salinity” (5 vs 10), “Type of salt” (artificial vs natural), “Female”, and “Stage”, is compiled in Table 4. Four-way ANOVA (nested design) revealed significant effects of all four factors on all biomass parameters, although at different levels of significance. Interactions between these factors were in all cases statistically significant. The developmental “Stage” had consistently the strongest influence on all biomass parameters per individual and on the percentage values of protein, C, N, and lipid, as well as the C/N ratio (all P < 0.001). The factor “Salinity” had generally stronger effects on biomass per individual and chemical composition (all P < 0.01) than the factor “Type of salt” (all P < 0.05).