A Multidisciplinary Experimental Study on the Effects of Breeders Diet on Newborn Seahorses (Hippocampus guttulatus)

The global trade in seahorses is increasing and new rearing techniques are currently available for a few species. One of the main bottlenecks in seahorse production is reproductive success, which is dependent upon the administration of high quality diets to breeders. However, little is known about the most adequate diets, on how they should be administered and on how they might affect newborn performance. In this study, three homogeneous groups of adult seahorses Hippocampus guttulatus were maintained in captivity. Each group was fed on one of the following diets since the onset of the breeding period: unenriched adult Artemia (diet A), unenriched adult Artemia and mysidaceans (diet AM) and mysidaceans (diet M). As a positive control, a wild pregnant male was used. The mixed diet AM provided the highest overall breeding success and reasonable fatty acid profiles, and better resembled newborn from a wild male. The use of mysidaceans solely (Diet M) provided the worse results in spite of the high DHA content. Significant differences were observed in newborn characteristics and fatty acid profiles among dietary groups and along the breeding season, especially in total n-3 HUFA and DHA. Monospecific diets including Artemia or mysidaceans very likely suffered nutritional deficiencies, which were reflected in morphological alterations of the muscle tissue (diet A) or abnormal early mortalities in newborn (diet M). Three progressive stages were identified along the whole breeding period: initial mixed capital-income period (100–120 days since the onset of the breeding period), followed by an inflection period (short transitional stage – income sources), and finally a long period characterized by the use of income sources and the progressive exhaustion of body reserves. Special attention should be deserved to the progressive changes in fatty acid profiles along the breeding season, resulting in a decrease in newborn performance. Considering the time required for a diet to be reflected in newborn, we recommend Hippocampus guttulatus breeders to be fed on a high quality diet for at least 3 months prior the breeding period.


Introduction
The global trade in seahorses is estimated to exceed 20 million animals per year to sustain the aquarium trade, the traditional Chinese medicine and as souvenirs (Vincent, 1996;Planas et al., 2017a). In addition, because of the destruction and degradation of their coastal habitats (seagrasses, coral reefs and mangroves), wild populations are progressively declining (Vincent, 1996;Martin-Smith et al., 2004). At present, many are the concerns over their long-term viability in nature (Vincent, 1996;Vincent et al., 2011, Kumaravel et al., 2012 and all seahorses species are included in Appendix II list of endangered species by CITES (Convention for the International Trade in Endangered Species).
Aquaculture represents a valid alternative to wild captured seahorses but ex-situ production still needs to face several biological and technical challenges (Cohen et al., 2016;Planas et al., 2017a). Both, low spawning quality and juvenile survival rates are the most critical factors for optimizing commercial seahorse aquaculture (Adams et al., 2001;Chang and Southgate, 2001;Lin et al., 2007). Reproduction represents a key aspect in breeding success, and broodstock feeding / nutrition is doubtless, one of the key factors to sustain reproduction in teleost (Migaud et al., 2013). The dietary effect on the output performance (egg and fry quality and quantity) is well known in shes, and many studies have reported the importance of lipids and fatty acids (Rainuzzo, 1993;Bromage, 1995;Izquierdo et al., 2001;Sargent et al., 2002;Tocher, 2003Tocher, , 2010Pavlov et al. 2004). Broodstock preparation is thus a critical step in the success of the spawning process of a certain species. On this regard, many studies have been directed towards egg quality in oviparous shes and the underlying characteristics to produce viable fry (Kjørsvik et al., 1990;Brooks et al., 1997). The egg's potential to produce viable fry is determined by several physical, genetic, and chemical factors, including the initial physiological processes occurring in the egg itself (Kjørsvik et al., 1990).
The ongoing process of oocytes maturation is one of the most metabolic demanding activities in sh and the quality of parental nutrition is thus of primary importance for the appropriate allocation of macromolecules into the oocytes and their maturation (Izquierdo et al., 2001;Volkhoff and London, 2018). However, there is a scarcity of studies focused on breeding performance and egg or fry quality in shes with male pregnancy such as syngnathids (pipe sh and seahorses). Due to the uncommon nature of reproduction, syngnathids offer a unique opportunity to examine pathways of repercussion of breeder's dietary resources on breeding performance and the resulting newborn quality. Reproduction in seahorses has received increasing attention in recent years, mostly on the effect of environment (e.g. photoperiod regime, temperature) (Lin et al., 2006(Lin et al., , 2008bPlanas et al., 2010Planas et al., , 2013, zootechnical conditions (e.g. sex ratio, aquaria design, seahorse size) (Woods et al., 2000;Dzyuba et al., 2006;Planas et al., 2008;Faleiro et al., 2008Faleiro et al., , 2016, and biological / physiological features (Boisseau, 1967;Carcupino et al., 2002;Poortenaar et al., 2004;Stölting and Wilson, 2007;Scobell and MacKenzie, 2011;Zhang et al., 2019; Wittington and Fresien, 2020). Only a few studies have focused on the effects of feeding and dietary sources on breeding success and newborn features in seahorses under ex-situ conditions (Lin et al., 2007;Faleiro and Narciso, 2010;Binh and Serrano, 2012;Saavedra et al., 2015;Palma et al., 2012Palma et al., , 2017Otero-Ferrer et al., 2012. Seahorses (Hippocampus spp.) are batch spawners with cyclical egg maturation and repeated mates within the breeding season. The eggs produced by females are transferred to the brood pouch of males, where embryogenesis will take place for 2-4 weeks, depending on the species and temperature (Foster and Vincent, 2004). The sealed brood pouch of the male works as a pseudo-placenta, not only providing protection and oxygen to the developing embryos, but also nutrients (Wilson et al., 2001;Carcupino et al., 2002;Stölting and Wilson, 2007). As a consequence both female and male nutrition play a role on the embryonic and fry quality and development, but unfortunately, this topic remains one of the less studied subjects in seahorses (Olivotto et al., 2011a;Cohen et al., 2016).
The European long-snouted seahorse Hippocampus guttulatus Cuvier, 1829 has been identi ed as a potential candidate for the marine ornamental trade (Cohen et al., 2016). Rearing techniques are currently available for this species but the breeding success (especially mating events) are still not optimized (Olivotto et al., 2011a;Planas et al., 2017a). Male competition is rather frequent resulting in mating interruption, and juvenile number and size decrease over the time in captive conditions (Faleiro et al., 2008). A bottleneck in breeding performance relies on both type and nutritional quality of diets. In captivity, both frozen and alive feed are usually used to nourish the broodstock, including mysid shrimps, amphipods and Artemia brine shrimps (Woods and Valentino, 2003;Olivotto et al., 2008;Palma et al., 2008;Murugan et al., 2009;Planas et al., 2017a). As concerns mysid shrimps, they are generally considered an optimal diet for seahorses Valentino, 2003, Otero-Ferrer et al., 2012).
Nutritionally enriched Artemia has received great attention and is presently widely employed (Woods, 2001;Wong and Benzie, 2003;Woods and Valentino, 2003;Planas et al., 2008), especially because it is easily cultured up to the adult stage, provides satisfactory n-3 HUFA and DHA contents after a long-time enrichment (Planas et al., 2017b), and supports high growth rates in H. guttulatus (Planas et al., 2008). However, it has been reported that the breeding performance in H. hippocampus decreased when fed on Artemia compared to diets including mysidaceans (Otero-Ferrer et al., 2012).
In the present study, the following questions have been addressed: (1) How experimental diets are assimilated (stable isotope analysis) by breeders and re ected in general features and biochemical composition of Hippocampus guttulatus newborn?, (2) Do mixed diets perform better than monospeci c ones?, and (3) Does H. guttulatus follow a mixed income-capital breeding strategy?. The main objectives of the study were (1) to propose a breeder's diet enhancing reproduction performance in the selected species, and (2) to ascertain when and how long the diet should be administered to provide its bene cial effect. To our knowledge, this is the rst multiapproach study on the effect of breeder's diet on reproductive performance and newborn quality in H. guttulatus.
Artemia cysts (AF, Inve, Spain) were hatched at 28 °C for 20 h in 20 L units, and the hatched nauplii gently rinsed with tapwater, collected on a 125μm mesh, rinsed and offered to seahorse juveniles.
Adult non-enriched Artemia was produced to feed adult seahorses during the breeding period. For that, the nauplii were hatched from MC450 cysts (Ocean Nutrition, Spain) and grown in 100 L units, at 26-28°C with gentle aeration and constant light. Adult Artemia was long-time enriched (3-6 days) to feed breeders prior to the start of the experiment but not during the experimental period. The enrichment was carried out in Artemia from day 16 onwards on a mixture consisting on live microalgae P. tricornutum and I. galbana (107 cells mL-1 ), Red Pepper (0.015 g L-1) and dried Spirulina (0.03 g L-1) (Planas et al., 2017).
Copepods (Acartia tonsa) were cultivated for the early feeding of seahorse juveniles in 700 L tanks at 26-27 °C and 38 salinity, at an initial density of 1 copepod mL-1. Copepods were fed every two days on the microalgae R. lens (103 cells mL-1). Siphoning of the culture tanks and water renewals (10% of the total volume) were carried out three times per week.

Seahorse Breeding and Rearing
Animal capture, handling and sampling were conducted in compliance with all bioethics standards on animal experimentation of the Spanish Government (Real Decreto 1201/2005, 10 th October 2005) and the Regional Government Xunta de Galicia (REGA ES360570202001/15/FUN/BIOL.AN/MPO01).
The adults of Hippocampus guttulatus seahorses used in the present study were reared in captivity or collected in Galicia (NW Spain) from August 2016 to January 2017 with permission of the Regional Government Xunta de Galicia. Prior to the start of the experiment, seahorses were maintained at the facilities of Instituto de Investigaciones Marinas (CSIC) in Vigo (Spain) in 320 L aquaria in a semi-closed system with a 10-14% daily renewal of seawater (Planas et al., 2008), and submitted to temperature and photoperiod natural-like regimes (Planas et al., 2010(Planas et al., , 2013, uctuating from 15 ºC (Winter) to 19 ºC (Summer) and from 10L:14D (Winter) to 16L:8D (Summer), respectively. Water quality was checked periodically for NO2, NO3 and NH4/NH3 content (0 mg L-1). Salinity and pH levels were 37 ± 2 and 8.0 ± 0.2, respectively for both species. Wastes and uneaten food were removed daily by siphoning the bottom of aquaria.
Males and females were maintained separate until the start of the experiment in mid-March 2017, when all seahorses were randomly distributed (1:1 sex ratio; 6 pairs per aquaria) in three 320 L aquaria and maintained for a whole breeding season (until early October 2017). All aquaria were maintained under the same conditions except for the diet administered. Temperature and photoperiod regimes were as described above. The average size (curved standard length) and weight of seahorse breeders (17.4 ± 2.6 cm and 11.50 ± 5.0 g, respectively) did not differ across experimental groups (ANOVA, p = 0.702, 17.4 ± 2.6 cm; ANOVA, p = 0.470, 11.5 ± 5.0 g). Seahorse broodstocks were continuously monitored to check for newborn release from male's pouch.
Prior to the breeding period, all breeders were fed for 2 -5 months on a diet based on enriched adult Artemia and captured/frozen mysidaceans (see below). During the breeding period, each husbandry aquaria received two daily doses (ad libitum) of different diets ( Figure 1 Diet AM: Mixture (1:1) of diets A and Small portions of the breeders dorsal n (partial n-clipping; about 20 m2 surface) were sampled at the onset and at the end of the breeding period, and properly stored for further stable isotopic analysis (SIA) .
Newly released juveniles were carefully collected by siphoning, counted and transferred (2-5 juveniles L-1) to 30 L pseudo-kreisel aquaria connected to a semi-opened recirculation system (Blanco et al., 2014). The newborn were reared until day 7 after male's pouch release (DAR) at 19 ºC under a constant 14L:10D photoperiod. The aquaria were lled with seawater ltered by a series of lter-cartridges (20, 10, 5 and 1 μm) and UV treated (76w; 16 L min-1) (JR1/50). The rearing system included a degasifying column and two 50 L chambers including mechanical (up to 20μm) and biological lters (perforated plastic bio-balls) and aerators. From the bio lter unit, the seawater was pumped to 36w UV units (AquaMedic®, Germany) and then to a 50 L reservoir aquarium, being nally routed by gravity towards the rearing aquaria .
Dead seahorses were daily removed (8:00 am and 15:00 pm) and counted. Final survivals of juveniles were recorded at 7 DAR (M50 at 18 -21 °C occurs at 6.3 -6.7 DAR) . For each experimental group, samples of newborn juveniles (see further sections for details) were randomly collected before rst feeding, euthanized with Tricaine MS-222 (0.1 mg L-1, Sigma Aldrich), rinsed with distilled water, and conserved at 80 °C for further fatty acids and isotopic analyses.
For length and weight measurements, sampled seahorses were transferred to Petri dishes, photographed and weighed individually on a Sartorius microbalance MC210P (± 0.01 mg). Curved standard lengths (SL = head + trunk + curved tail) were measured from digital photographs using an image processing software (NIS Elements, Nikon).

Sample Analysis
Total lipids from live prey and juveniles (10-20 mg dry weight per sample, in duplicate) were extracted according to Bligh and Dyer (1959). Aliquots of total lipid extracts with known lipid content were centrifuged, resuspended with 0.5 M ammonium formate solution, freeze-dried and stored at -80 °C until further fatty acids analyses. Total lipid content was quanti ed gravimetrically (Herbes and Allen, 1983). Fatty acid (FA) composition of lipids was analyzed by gas-chromatography (GC) according to Christie (1982). Lipids were transmethylated (Lepage and Roy, 1986) and fatty acids analyzed by GC (Perkin Elmer, Clarus 500 gas chromatograph) as described in Planas et al. (2010). Samples were analyzed in duplicate.
Carbon (δ13C) and nitrogen (δ15N) isotope signatures and elemental composition (total C and N) were analyzed in both prey and seahorses samples ( n of breeders and bulk juveniles). Frozen juveniles (n > 5; pooled individuals) were dried for 48 h (60 ºC), weighted and manually homogenized. The analyses were made on sub-samples of 0.2-1.0 mg dry weight biomass at Servizos de Apoio á Investigación (SAI) of the University of A Coruña (Spain) as reported by Valladares et al. (2016). As defatting of samples prior to isotopic analysis is recommended when lipid content exceeds 5% weight (C:N >3.56) (Post et al., 2007), samples were not defatted but arithmetical corrected factors (Supplementary Table 1) were applied considering defatting (copepods, Artemia nauplii and metanauplii, and seahorse ns or juveniles) or defatting + acidi cation (10% HCl) (Mysidacea and adult Artemia). The samples were measured by continuous ow isotope ratio mass spectrometry using a FlashEA1112 elemental analyzer (Thermo Finnigan, Italy) coupled to a Delta Plus mass spectrometer (FinniganMat, Bremen, Germany) through a Con o II interface. Stable isotope abundances are expressed as permil (‰) in conventional delta relative to VPDB (Vienna Pee Dee Belemnite) and Atmospheric Air. The precision (standard deviation) for the analysis of δ13C and δ15N of the laboratory standard (acetanilide) was ± 0.15‰ (1-sigma, n=10). Standards were run every 10 biological samples.
Transmission electron microscopy (TEM) was applied to abdominal portions of seahorses (three newborn per dietary treatment). The samples were processed for Durcupan ACM (Fluka) resin embedding as follows: pieces xed in 2.5 % glutaraldehyde in 0.2 M phosphate-buffered saline (pH 7.4), for 2 h at 4°C , were washed repeatedly in the same buffer and post xed in 1 % osmium tetroxide in 0.2 M phosphatebuffered saline for 1 h at 4 °C. Then, the tissues were dehydrated with increasing alcohol concentrations. The dehydrated pieces were embedded in Durcupan ACM (Squadrito et al., 2017). Finally, the sections were obtained with a Reichert Jung Ultracut E. Semi-thin sections (1 μm) were stained with toluidine blue and examined with a light microscope. Ultrathin sections (740 Å) were obtained from selected areas of the semi-thin sections, stained with uranyl acetate and lead citrate, and examined and photographed with a transmission electron microscope (JEOL JEM 1400 Flash) (Viña et al., 2014).

Data Analysis
All statistical analyses were conducted in R v.3.6.1 (R Core Team 2019). Variability across treatments was examined by ANOVA or non-parametric Kruskal-Wallis test for independent samples, depending on signi cant deviations from normality and homogeneity of variance, as examined by the Shapiro-Wilks and Levene tests, respectively (ggpubr v0.2.5 R package) (Kassambara, 2020a). When signi cant, A Kruskal-Wallis test followed by Wilcoxon range comparisons were applied (Pgirmess v1.6.9 package of R) (Giraudoux et al, 2018). Significance levels were set at p<0.05.
The Jonckheere-Terpstra trend test (J-T test) was applied to individual FA data for an ordered alternative hypothesis within an independent samples (between-seahorse groups) design (clinfun v1.0.15 package in R) (Venkatraman, 2018). The test has more statistical power than the Kruskal-Wallis test, incorporating information about whether the order of the groups is meaningful (Siegel and Castellan, 1988). Hence, we applied the test to check whether the more proportions of mysidaceans in the diet (order: A -AM -M), the more the independent variable tested will go down or up.
The incorporation of the experimental diets in breeder tissues was analyzed considering dietary isotopic pro les (Table 1) and the resulting isotopic changes in dorsal ns (initial vs nal) ( Table 2). Differences in isotopic signatures of diets corresponded approximatively to one trophic level, with values ranging from -22.12 to -20.53 ‰ for δ13C (Δ = 1.59) and from 7.67 to 10.83 ‰ in δ15N (Δ = 3.16).
Initial isotope values for δ13C and δ15N in dorsal n at the onset of the experimental breeding period did not differ signi cantly (ANOVA, P = 0.192 and 0.143, respectively) ( Table 2). At the end of the experiment, changes in signatures for both isotopes in ns were inversely correlated (Linear regression, R2 = 0.994) across treatments (Supplementary Figure 2) and strongly agreed with isotope values of the corresponding diets, with signi cantly lower δ15N signatures in the group fed uniquely on Artemia (K-W test, P= 0.044). Isotopic changes (Δδ) in ns during the breeding season were 2.97, -0.20 and -1.89 for δ13C, and -1.42, 0.88 and 2.58 for δ15N, respectively (Table 3). Isotopic changes between nal isotopic values in dorsal ns and diets A, AM and M were 8.13, 4.14 and 2.50 for δ13C, and 2.62, 4.25 and 3.91 for δ15N, respectively.

Breeding performance
The total amount of newborn batches produced in treatments A, AM and M were 4, 7 and 3, respectively, and the total production of newborn were 534, 1416 and 248, respectively (Table 4). Average batch size in treatment AM (202 ± 82 newborn; max = 343) was higher than in treatments A (134 ± 83) and M (84 ± 72) but differences across treatments were not signi cant (ANOVA, P=0.331) due to the large standard deviations of means. Newborn production started earlier and extended over a longer period for group AM breeders (from day 72 to 210 of the breeding period). The opposite occurred in group M (from days 152 to 179). In group A, newborn batches were released from days 121 to 202.
Survivals at 7 DAR were high but not signi cantly different across treatments (K-W test, P=0.066) (Table  4), ranging from 94.0 ± 5.3 % in treatment A to 61.4 ± 19.2 % in treatment M. First mortalities in treatment M occurred at 2 DAR, with a peak at 4-5 DAR (Figure 2). Similarly to newborn reared from a batch released by a wild male, rst mortalities in treatments A and AM were delayed until 3-4 DAR.

Newborn characteristics
Pooled samples of newborn differed in δ15N across treatments (K-W test, P = 0.044) but not in δ13C, ranging from 11.21 to 14.96 ‰ and from -17.31 to -13.86 ‰, respectively (Table 2; Supplementary  Figure 3). Mean δ15N value in group A was signi cantly lower than those from the other groups.
However, mean isotopic values did not re ect the pattern of change across treatments during the breeding period. The low number of batches released in treatments A and M did not allow a precise analysis of isotopic changes but a trend could be inferred from treatment AM, in which a progressive change in δ15N towards those in the diet was noticed throughout the breeding season, with signals decreasing from 16.0 to 12.3 ‰.

Global Assessment
From the correlation network developed for treatment AM (Figure 3) it can be observed that time (days elapsed from the onset of the breeding season) negatively affected survival, the relative content in saturated FA (particularly 14:0) and δ15N values, but increased the content in total n-6 FA (mainly 18:2n-6c) and 18:3n-3. Length and dry weight of newborn were positively correlated but they did not correlate with time. Size and weight were negatively correlated with δ13C. In addition, weight was negatively and positively correlated with total monoenes and 18:0, respectively. The full results on correlation analyses are provided in Supplementary Figures 4 and 5. Sample similarities were remarkably high (R2>0.7).
However, early released newborn batches from group A were more similar to intermediate batches from group AM, and late released batches were similar, independently of the group considered. The only batch available by a wild male was similar to early batches released by males fed on diet AM.

Transmission Electron Microscopy
Ultrathin sections of abdominal portions of newborn seahorses were observed by transmission electron microscopy (TEM) in order to analyze the ultrastructural features of muscular tissue. Generally, TEM micrographs of skeletal muscle sections revealed a regular morphology with a normal arrangement of sarcomeres, resulting in a regularly striated appearance with a normal alternation of dark anisotrope band A (myosin and actin) and light isotrope band I (actin lament) ( Figures 5A, B, D). In the longitudinal sections of TEM micrographs of skeletal muscle sections, the myo brils appeared tightly packed in the scarce amount of cytoplasm. The nuclei were located in the peripheral layer of the bre and mitochondria beneath the sarcolemma and among the myo brils. Besides, the regular repeating pattern of myo laments in the myo brils were detected displaying dense Z lines, forming a regular zigzag ( Figures  5A, B, D) at the center of the light I bands. The A bands showed a central light H zone divided by a dark M line ( Figure 5A). In addition, the organization of the sarcoplasmic reticulum and the sarcoplasmic triad, consisting of two terminal cisterns and the transverse tubule T (formed by the invagination of the sarcolemma) were also observed ( Figures 5 A, C). Scattered satellite cells surrounded by volumetrically modest cytoplasm were often seen in muscle tissue (Figures 5 E, 6B).
In the cross sections, the disposition of the two sets of myo laments in different bands of the sarcomere were analysed. The myo brils were well delineated by the close apposition of the sarcoplasmic elements (Figures 6 A, B). The nuclei were located in the peripheral layer of the bre and mitochondria beneath the sarcolemma and among the myo brils.
The general morphology observed in newborn released by males from diet A appeared similar to the other groups (M and AM). However, some ultrastructural muscle changes were noticed. In particular, group A displayed regressive changes in some muscle bers characterized by a disarrangement of myo brils ( Figure 5B), with evident myo brillary interstitial spaces. Besides, a discontinuity of the sarcolemma and myelin-like bodies were present ( Figure 6C). In addition, a disorganization of the sarcoplasmic elements and of the invagination of the sarcolemma (i.e., tubules T) was also noticed. An interruption of the Z line was also occasionally evidenced (Figures 5 E, F) and neuromuscular junctions were observed between motor nerve endings and muscle bers. Particularly, the terminal axons of the motor neuron were spherical or oval in shape and t into smooth depressions in each muscle bre ( Figures 6D-F). At the presynaptic zone, several mitochondria and small clear synaptic vesicles were noticed. Other morphological alteration were not detected at the level of the above structures.

Discussion
Breeding performance In this study, the isotopic results achieved in dorsal ns of the seahorse Hippocampus guttulatus con rmed the effective assimilation of experimental diets by breeders, with effects on average isotopic signatures and fatty acid pro les in newborn seahorses across groups. The overall breeding performance (extension of the effective breeding period, batches produced, newborn released) resulted enhanced in parents fed on the mixed diet AM (non-enriched Artemia and mysidaceans), suggesting a better nutritional condition than in groups A and M. Unexpectedly, the use of mysidaceans as only prey (Diet M) resulted in the worse breeding success among the three experimental diets.
Brood size in some seahorse species (e.g. H. kuda, H. erectus) has been related to the length of brood pouches of males and trunk of females (Dzyuba et al., 2006;Lin et al., 2012). The newborn size might be correlated to brood size, gestation time and nutrient supply in the same males and females' investment (Lin et al., 2008b(Lin et al., , 2012. Besides this, larger H. erectus males (larger brood pouches) do not produce larger broods, and male size can be positively correlated with juvenile size (Lin et al., 2012). In H. guttulatus, it has been pointed out that the size of newborn seahorses can be, in part, paternally determined (limited carrying capacity of the male), as male size is negatively correlated with embryo density and positively correlated with newborn size (Faleiro et al., 2016). As breeders average size did not differ across our experimental groups, we assume that differences between treatments in our study were due to dietary dissimilarities. Differences in the capability of seahorses to digest and assimilate a given prey is another subject to consider when assessing diet quality. The monitoring of prey in faeces of H. guttulatus by 18SrDNA ampli cation showed that gut passage times of crustaceans (Artemia, Palaemonetes, and mysidaceans Leptomysis sp. and Siriella sp.) are species-dependent (longer passages times in mysidaceans), probably due to differences in their digestibility (lower digestibility in Palaemonetes) (Corse et al., 2015).
Mysidaceans have been considered an excellent prey for the on-growing and breeding of some seahorse size) and newborn size in breeders fed on mysids, presumptively due to a higher nutritional quality of mysidaceans (basically in DHA content). Those results strongly agree with those with diet AM, but not with the poor results achieved with diet M. The authors also recommended (but not tested) the use of a mixed diet including Artemia and mysids.
The survivals achieved in newborn from group A were unexpectedly high, being accompanied by small initial lengths and reduced batch sizes. Interestingly, it has been reported in H. reidi that reducing the nutritional quality of a male's diet since one month before conception resulted in abnormally large and heavy newborns with extremely reduced survivals, independently of the dietary quality in females . These ndings partially agree with those reported in the pipe sh Syngnathus typhle, in which a higher nutritional condition in males brooding heavier embryos was correlated with a higher embryo survival (Sagebakken et al., 2017).
Premature newborn or egg clutch dropping were not observed, contrarily to previous ndings in H. hippocampus fed on Artemia (Otero-Ferrer et al., 2012) or commercial mysidaceans (i.e. much lower DHA and HUFA levels than in the present study) (Saavedra et al., 2015), and in H. guttulatus fed on suboptimal quality of Artemia (i.e. high lipid and low DHA content) (Planas et al., 2009a(Planas et al., , 2010. In the later, the issue was solved by optimizing DHA and EPA content in Artemia (Planas et al., 2017b) and isolating pregnant males a few days prior to newborn release (Olivotto et al., 2011a). Even though the growth of adult H. guttulatus fed on Artemia may be satisfactory (Planas et al., 2008(Planas et al., , 2009bSegade et al., 2015), it seems that its high lipid content and unbalance fatty acids pro le might result in an excessive accumulation of lipids in the liver and an abnormal ( for H. guttulatus adults in our nearby region (NW Iberian Peninsula) are amphipoda (caprellids and gammarids), whereas mysidaceans are insigni cant prey in the bulk diet composition (Valladares et al., 2016). We recently demonstrated that diets including mysidaceans (alone or mixed with Artemia) provide the best trophic enrichment factors (laboratory derived) to infer the composition of bulk diet in wild syngnathids by means of isotopic signatures (Planas et al., 2020). Hence, even though mysidaceans might be considered a high quality prey for some seahorse species, it is very likely not suitable for breeding H. guttulatus when administered as the sole prey. Unfortunately, nutritional requirements in seahorses are unknown, and provision of suitable food sources for ex-situ culture often represents a serious constraint (Chang and Southgate, 2001) as not all species seem to have the same nutritional needs, as suggested by inter-speci c differences in FA pro les (Lin et al., 2008a). Consequently, mixed diets (present study) and/or dietary supplementation would potentially reduce some nutritional de ciencies of single prey (Palma et al., 2012(Palma et al., , 2017.

Breeding strategy
Seahorses are synchronous batch spawners, as several oocyte populations are present during the reproductive season and clutches of oocytes develop, mature, and are released simultaneously (Fernández-Palacios et al., 2011). During the last part of the secondary growth of oocytes (vitellogenesis), lipids and yolk proteins are incorporated and stored in large amounts in the yolk. That process and further embryogenesis of fertilized eggs in males cyclically occurs along the breeding season, requiring a continuous expenditure of energy and nutrients. Seahorse females and males are time-synchronized for egg batch transfer from females to males and newborn release in males, and in H. guttulatus kept at 19 -20 °C both processes occur at 25 -27 days time-intervals (Planas et al., 2010). On this regard, it is very important to ascertain the origin (capital, income or mixed capital-income reproduction strategy) of the energy and nutrients necessary for reproduction (Williams et al., 2017), and to evaluate whether and when egg/fry composition can be affected by the type of diet provided. Hence, the identi cation of both diet quality and effective feeding periods (the time period needed to change the biochemical composition of eggs/fry) for optimum broodstock performance is of pivotal importance in enhancing newborn quality (Fernández-Palacios et al., 2011;Griffen, 2018). The present study and the pattern of isotopic changes discussed elsewhere (Planas et al., 2020) suggest that the reproduction type followed by H. guttulatus corresponds to a mixed capital-income strategy. This strategy usually implies a progressive drop in newborn quality in successive batches (McBride et al., 2016, Griffen, 2018. In order to properly assess the breeding strategy followed by H. guttulatus, the effects of experimental diets must be analyzed over the entire breeding period. Given that the pattern of change in the biochemical components analyzed in newborn did not follow a mathematical model for tting, the effective feeding periods were inferred from the smoothed trends in the most remarkable components along the breeding period (particularly for group AM) (Figure 7). In most cases, trend in exions occurred within the period comprising 100 and 140 days, depending on the variable and seahorse group considered. We identi ed three stages along the whole breeding period: Stage 1 (mixed capital-income sources): Progressive changes occur since the dietary shift at the onset of the breeding season. Energy and nutritional resources would mostly originate from resources gained prior to the breeding period (capital breeding), accompanied by an increasing use of sources derived from the experimental diet (income breeding). This stage would extend until days 100 -140 since the onset of the breeding In exion period (income sources): This is a transitory period indicating the end of the capital-income breeding strategy and the use exclusively of resources from the experimental diet. In this period, differences among treatment were clearly Stage 2 (income sources and exhaustion of body reserves): In most cases, differences among groups increased or decreased (depending on the variable considered) beyond the in exion period.
Maximal divergences between treatments were reached during this period, notably by the end of the stage. The trend followed by certain variables (e.g. DHA, DHA/EPA) in late newborn batch releases suggests a more or less intense exhaustion of body resources, accompanied by decreases in survivals at 7 DAR and newborn size. This statement is also supported by the trends in newborn isotopic signatures (Planas et , 2020).
Fry quality can be manipulated to a certain extend by modifying the nutritional composition of the diet during the breeding season of shes (Fernández-Palacios et al., 2011), including seahorses (Otero-Ferrer et al., 2020). However, it is also important to assess the effect of preconception diet on reproduction success and the contribution of males and females, as reported by  in H. hippocampus. According to our above hypothesis on the stages recognized along the breeding season, the diet provided preceding the reproduction season would contribute for a certain period to the initial composition and quality of H. guttulatus newborn. Hence, the initial quality of eggs/newborn could be improved by feeding the breeders on a high quality diet for at least 3 months in advance of the onset of the breeding season, which is in agreement with previous ndings in the tropical species H. reidi (Planas and Chamorro, 2019). That nding is of primary importance to female diets as it is noteworthy that embryos are mostly dependent on maternal nutrition (yolk quality) until later stages of development, when the embryos are mostly dependent on paternal contribution (Otero-Ferrer et al., 2020).

Newborn composition
Breeders diet must be optimized to ensure good early development (Izquierdo et al., 2001;Migaud et al., 2013). Currently, nutrients requirements are rather well known in many marine species with interest in ornamental trade, and optimal dietary formulations are available (Craig et al., 2017). However, there are two main issues in the feeding of seahorses in captivity: the lack of knowledge on their nutritional needs and the rejection of inert formulated diets (Olivotto et al., 2011a;Palma et al., 2012;Planas et al., 2017a).
As the broodstock diet is strongly re ected in the composition of eggs/fry, particularly for lipids (Wiegand, 1996), the composition of eggs or fry can be used as a proxy to infer nutritional requirements in shes to improve breeding success (Sargent et al., 1989Fernández-Palacios et al., 1995;Rainuzzo, 1993;Rainuzzo et al., 1997). The content in lipids and FA pro les in eggs of H. guttulatus under different dietary regimes were reported by Planas et al. (2008Planas et al. ( , 2010 and Faleiro and Narciso (2010). Essentially, FA pro les in eggs and newborn do not differ signi cantly when the content is referred to relative percentage of total FA, but they differ markedly when the content is given as dry weight percentage (Faleiro and Narciso, 2010). In the present study, we focused exclusively on newborn, including a batch released by a wild pregnant male for comparative purposes.
The cluster analysis (Supplementary Figure 5) and PCA ( Figure 4C) performed on newborn revealed a higher similarity between the wild batch (released early in de season) and the early and mid-term batches released from group AM. The similarity among those batches was mainly due to DHA, total polyunsaturated FA, δ15N, survival at 7 DAR, DHA/EPA ratio, AA and, to a lesser extent, to minority 20:1n-9 and 20:2n-6 (<1 %). The DHA/EPA ratio re ected the higher and more stable content in DHA thorough the breeding season, compared to treatment A. The essentiality of DHA, EPA and AA in early developmental stages of marine sh has been previously discussed in depth (Rainuzzo, 1993;Bromage 1995;Wiegand, 1996;Rainuzzo et al., 1997;Estévez et al., 1999;Tocher, 2010). There is no doubt about the nutritional essentiality of certain HUFAs in marine sh. These FA are important sources of energy in sh during early embryogenesis , when DHA and AA are highly conserved in most teleost (Watanabe et al., 1989). However, Faleiro and Narciso (2010) analyzed the lipid and FA dynamics during early development of H. guttulatus (breeders fed on adult Artemia; Artemia diet not provided) and reported that EPA, DHA and palmitic acid were not preferentially retained in embryos as they constitute the main energy sources, with EPA being catabolized at a high rate. Consequently, an exceptionally low lipid content (5.0 % DW) and extremely low EPA levels (5.0 %) were reported in newborn. Compared to that study, EPA content and total lipids in newborn from our study were 2-fold higher, in agreement with previous results in freshly released eggs (Planas et al., 2008(Planas et al., , 2010. In spite of the extremely low levels of DHA in diet A, its average content in newborns was high (with a sharp decrease in late batches) although lower than in groups AM and M. This nding indicates the essentiality of DHA in seahorses. Overall, FA pro le and DHA/EPA ratio in newborns released from the wild male showed a higher similarity with newborns released by males fed on diets including mysidaceans. Hence, the variable effect of breeder's diet on the allocation of FA resources in newborn from different experimental assays is con rmed in H. guttulatus, deserving further attention, especially with regard to dietary DHA requirements and further effects on survival and size of newborn.
Higher dietary contents in total n-3 HUFA and DHA does not necessarily imply higher quality. Excessive levels of dietary n-3 HUFA can reduce eggs production and survival, and cause yolk sac hypertrophy in sh larvae (Fernández-Palacios et al., 1995). As the TEM did not show any morphological alteration in tissues of newborn from diet M, we suspect that n-3 HUFA and DHA levels in diet M were below detrimental levels, and that the low breeding success in that treatment was due in part to energetic and/or nutritional dietary de ciencies. It is known that egg viability may decrease in lower quality diets (Saavedra et al., 2015), which is in agreement with the reduced newborn production in diets A and M. As eggs dropping did not occur in our study and female-male egg transfers were not checked, viability in eggs transferred to males is not available. Hence, the reasons for that low performance are uncertain. In addition to potential negative effects of low quality diets on mating displays, we did not discard the occurrence in seahorses of the strategy of cryptic choice reported in other syngnathids, in which males increase rates of offspring abortion in pregnancies to absorb nutrients resources for future reproductive opportunities (Paczolt and Jones, 2010;Sagebakken et al., 2010).

Conclusions
In the experimental study presented here, isotopic signatures of experimental diets supplied during the breeding season were re ected in breeder's dorsal ns (e cient assimilation of diets) and subsequently in newborn, with higher δ15N signals in breeders and newborn from the diets including mysidaceans. Monospeci c diets including Artemia or mysidaceans very likely suffered from nutritional de ciencies, re ected in morphological alterations of the muscle tissue or abnormal early mortalities in newborn. Those de ciencies seemed to be overcome with the mixed diet AM (unenriched Artemia and mysidaceans), which provided the highest overall breeding success and reasonable fatty acid pro les, and better resembled newborn from a wild male. However, special attention deserve the progressive changes in fatty acid pro les along the breeding season, resulting in a decrease in newborn performance.
Considering the time required for a diet to be re ected in newborn, we recommend that Hippocampus guttulatus breeders be fed on a high quality diet at least 3 months in advance of the onset of the breeding period. In order to further optimize captive breeding and rearing of seahorses, further research should be addressed on wild populations with particular emphasis on broodstock feeding habits, on how feeding preferences possibly change during the breeding season and on potential shifts depending on the geographical distribution of the sh. Data collected from the wild will help scientist in the selection of the more suitable diet able to sustain the best egg and fry quality. Tables   Table 1. Average isotopic pro les (δ13C ‰ and δ15N ‰), total C and N, and C/N ratios in prey (A -Artemia, Mc -Captured mysidaceans, Mf -Frozen mysidaceans) and in diets for breeders (Diets A, AM and M) of Hippocampus guttulatus. See text for further details on the composition of diets. Raw isotopic data were transformed considering conversions provided in Table 1 Declarations Con ict of Interest Statement The authors declare that the research was conducted in the absence of any commercial or nancial relationships that could be construed as a potential con ict of interest.

Data Availability Statement
The datasets analyzed for this study can be found in the Pangaea collection (doi pending). Data not listed in this collection is available from the authors upon reasonable request.

AUTHOR CONTRIBUTIONS
MP as leading author has done the conception of the work, breeding and rearing, sampling, lipid and stable isotope analysis, interpretation of data, and most part in drafting. IO and RL assisted in paper writing. MJG was responsible for fatty acid analysis. RL and MZ were responsible for TEM and this part of the draft. All authors revised the manuscript.

ACKNOWLEDGMENTS
We acknowledge support of the publication fee by the CSIC Open Access Publication Support Initiative through its Unit of Information Resources for Research (URICI). We are grateful to Alexandro Chamorro, Alex Paltrinieri, David Costas, Isabel Medina, and the staff of the Aquaria Service at IIM-CSIC for assistance in the maintenance of seahorses in captivity and technical support. The authors are also grateful to the Regional Government Xunta de Galicia for providing the permission for the capture of wild adult seahorses Hippocampus guttulatus. Finally, we are very grateful to Sergio Picazo (https://sergiopicazo.com/) for providing usage rights of seahorse illustration in gure 1. Figure 1 General scheme of the experimental framework (breeders fed on diets A, AM and M) and sampling (dorsal n and newborn). Seahorse illustration© by Sergio Picazo Ferro (https://sergiopicazo.com/).

Figure 2
Page 36/40 Initial survival of newborn seahorses released by a wild male and by captive males fed on diets A, AM and M.  Page 38/40 FA, n-3/n-6 (n3.n6), DHA/EPA, DHA/AA, and individual FA. The ellipse corresponding to 95% con dence or group AM is shown in yellow shadow. Other ellipses were not computed due to insu cient data.