Fertilized eggs of ABFT were obtained on 24 June 2013 from a spontaneous spawning of broodstock in the farming facilities at El Gorguel, Cartagena (SE Spain), of the Caladeros del Mediterráneo Company. The fertilized eggs were transferred to facilities of the Spanish Institute of Oceanography (IEO) in Mazarrón (SE Spain). An experimental rearing was undertaken with 50000 ABFT larvae. Larvae 2 days after hatching (DAH) were placed in a 5000-L tank. Larvae were fed on rotifers from the beginning to 18 DAH. Artemia salina was also supplied to larvae from 16 to 20 DAH. A piscivorous diet was started on 17 DAH, consisting of yolk-sac larvae of gilthead seabream (Sparus aurata), and between 17 and 20 DAH both types of diets were used to feed ABFT larvae.

The water temperature throughout the experiment was 24°C on average. The light regime was 14 hours per day and intensity was between 500 and 800 lx. The larval rearing experience followed the methodology described by Ortega et al. (2011) and De la Gándara et al. (2012). The experiment was performed according to Spanish law 6/2013 based on the European Union directive on animal welfare (Directive 2010/63/EU on the protection of animals used for scientific purposes).

ABFT eggs and hatched larvae were collected regularly every 2 days until 15 DAH. After this period, larvae were sampled every 3 days until 32 DAH. The larvae were anaesthetized using 100-200 mg l^–1 MS222 (tricaine methane sulfonate, Sigma-Aldrich, USA) depending on fish size. ABFT larvae and preys were immediately conserved in a –80°C freezer until laboratory analysis. The samples of ABFT larvae were measured for standard length (SL) with the Image-Pro Analysis Software. Afterwards, samples were dehydrated in a freeze dryer for 24 h and weighed with a precision microbalance (0.01 mg) for quantifying the relative composition of nitrogen and carbon isotopes.

Egg and larval samples were pooled for each age class until 6 DAH in order to reach the required dry weight (DW) for SIA analysis (1-2 mg). The larvae from 8 to 32 DAH were individually analysed. For SIA, aliquots of 7-10 replicates for each age class were analysed. The ABFT and prey samples were homogenized and packed in tin capsules after dry freezing. The content of ¹³C and ¹⁵N was measured by continuous gas flow system using a Thermo Finnigan Elementary Analyzer Flash EA 1112 coupled to a Finnigan MAT Delta Plus mass spectrometer at the Instrumental unit of Analysis of the University or A Coruña. Ratios of ¹⁵N/¹⁴N and ¹³C/¹²C were expressed in conventional delta notation (δ), relative to the international standard, atmospheric air (N₂) and Vienna Pee Dee Belemnite, respectively, using acetanilide as standard. The relative abundances of ¹³C and ¹⁵N were reported as isotopic ratios (parts per thousand, ‰) relative to standards.

For carbon SIA, a previous chemical extraction of lipids is required. This analytical procedure could not be carried out due to the low larval weight. Nevertheless, a posterior correction of the δ¹³C values for lipid content in ABFT larvae and preys was estimated using the model proposed by Logan et al. (2008). This empirical lipid correction has been previously used in fish larvae (Laiz-Carrión et al. 2013, 2015, Quintanilla et al. 2015).

The discrimination factor (Δ) of stable isotopes is the difference in the isotopic composition between the animal and its food source (Caut et al. 2009). This factor is influenced by the maternal isotopic signature in early life stages (δX_mother=δX_egg), since isotopic values of eggs and endogenous feeding larvae are totally determined by maternal transfer (Le Bourg et al. 2014). Therefore, this study estimated the isotopic discrimination factor by maternal transference in pre-flexion larvae, through Equation (1):

where δX represents the stable isotopic ratio (carbon or nitrogen) of ABFT eggs and pre-flexion larvae. The carbon and nitrogen stable isotopic ratio discrimination for maternal transference (ΔX_m) was estimated by the difference between the isotopic values of eggs and the pre-flexion larvae.

The estimated isotopic signature of maternal origin was calculated by solving Equation (2):

where δX_mother represents the stable isotopic ratio (carbon or nitrogen) of maternal origin through SL of pre-flexion larvae.

The discrimination factor by diet (ΔX) of reared post-flexion ABFT larvae was calculated when reaching steady state (24-32 DAH) with the diet, through Equation (3):

where δX represents the stable isotopic ratio (carbon or nitrogen) of the ABFT post-flexion larvae and δX_prey corresponds to the isotopic ratio of seabream yolk-sac larvae.

Nitrogen turnover rate was calculated in ABFT post-flexion larvae after diet transition over time until steady state was established between the larvae and its diet. These values were fitted to an exponential model (Eq. 4). The model was first applied by Hesslein et al. (1993) to tissue changes in larval whitefish (Coregonus nasus) and later by Gamboa-Delgado et al. (2008) to larvae of Solea senegalensis and by Tanaka et al. (2010) to larvae of Thunnus orientalis. The model integrates isotopic values with time, as well as their growth and metabolic rates through Equation (4):

where N_larvae is nitrogen isotopic value of larvae (whole body mass) at time t,
N_o is the nitrogen isotopic value of each larva in steady state with the initial diet,
N_n is the isotopic value reached when the larvae is in equilibrium with a new diet,
k is the growth rate expressed per day (d^–1), and m is the metabolic turnover rate (d^–1) that is derived from Equation 4 (Hesslein et al. 1993).

An estimate of the growth rate was derived from the exponential growth equation for weight (Eq. 5)

where W is the final DW, W₀ is the initial weight and t is the time (days).

Turnover rates can be measured by monitoring larval isotopic values over time until a steady state between larvae and their diet is reached. A function is then fitted to the change in isotopic composition over time, and can provide an estimation of the time necessary for half of the larvae to reach isotopic steady state after consuming a new diet. The half-life for nitrogen isotopes in post-flexion BFT larvae can be calculated (half time, t₅₀) (Eq. 6):

where t₅₀ is the half-life time of isotopic incorporation into larval tissue, and m and k correspond to variables cited above.

The Kruskal-Wallis test was applied to analyse the statistical differences of isotopic values of δ¹⁵N and δ¹³C during ontogenetic development until the early juvenile phase of ABFT when the isotopic discrimination factor reached isotopic steady state with prey. The maternal influence was estimated through to the linear correlation of the isotopic values of pre-flexion larvae minus the isotopic values of the eggs. All statistical tests were undertaken using the STATISTICAL 7.0 package (Statsoft) and the significance level for all analyses was set at α=0.05.

A total of 115 samples from reared ABFT were analysed to estimate the SIA values (δ¹⁵N, δ¹³C) in this experimental study, starting from the egg stage (–1 DAH) (n=10 pooled), to pre-flexion larvae (0-6 DAH, n=35 pooled; 6-12 DAH, n=28 larvae) and to post-flexion (15 DAH, n=42 larvae). The newly hatched larvae measured on average 3.5±0.23 mm SL and their mean DW was 0.07±0.00 mg. Yolk sacs of the larvae were fully consumed at 2 DAH. The relative growth relationship indicates slow growth during the pre-flexion phase (0-10 DAH) (Fig. 1). The notochord flexion started at 15 DAH, when larvae ranged between 5.5 and 6.5 mm SL. After 22 DAH, larvae showed fast growth in length and an exponential increase in weight with age (Fig. 2).

The isotopic composition of carbon and nitrogen showed differences during larval development throughout the rearing experiment (Kruskal-Wallis p<0.01) (Fig. 3), from the egg (–1) to 32 DAH. The stable isotopes of nitrogen and carbon ranged from 5.0‰ to 14.3‰ for δ¹⁵N and from –18.9‰ to –12.1‰ for δ¹³C.

Maximum δ¹⁵N values were found in the egg and yolk-sac stages, where δ¹⁵N values averaged 12.6±1.0‰ (n=40), (Fig. 3). Thereafter, δ¹⁵N showed a decreasing trend during the early stages of development until minimum values were reached at 15 DAH, coinciding with the initiation of notochord flexion. At 15 DAH, 85% of the larvae showed notochord flexion, corresponding to the lowest isotopic values of nitrogen (5.0±0.11‰). After flexion, δ¹⁵N values increased as larvae developed into early juveniles. From 24 to 32 DAH, δ¹⁵N values were stabilized at 13±0.34 ‰ (Fig. 3).

The isotopic composition of carbon showed an inverse trend to that of the stable isotope of nitrogen (Fig. 3). The lowest δ¹³C values coincided with the egg and yolk-sac stages. Thereafter, the carbon stable isotope increased steadily until 15 DAH, when the highest δ¹³C values were recorded (–12.3±0.1‰). Afterwards, δ¹³C values decreased steadily. Unlike the values of δ¹⁵N, those of δ¹³C did not reach a point of steady state between 24 to 32 DAH.

Our results showed the variation of the isotopic values during the three stages of larval development of ABFT (Fig. 3). We observed significant differences between different larval developmental stages, particularly between pre-flexion and flexion (Kruskal-Wallis, p<0.05), and between flexion and post-flexion (Kruskal-Wallis, p<0.05). Nonetheless, nitrogen and carbon isotopic values of the pre-flexion phase, containing isotopes of maternal origin, did not differ significantly from those of the ensuing post-flexion developmental phase (Kruskal-Wallis, p>0.05), owing to the high isotopic signature of the seabream yolk-sac larvae supplied as preys.

The highest δ¹⁵N and lowest δ¹³C values were observed in the eggs (-1 DAH) and the early pre-flexion larvae (0-4 DAH). The stable isotopes followed a time-line pattern with ontogeny. Therefore, it is logical to assume that the initial values of stable isotopes correspond to maternal transmission.

While δ¹⁵N started off with the peak values at the egg stage, δ¹³C values started at minimum levels. Both stable isotopes reached their turning point at 15 DAH (Fig. 3), coinciding with the initiation of notochord flexion.

The isotopic discrimination factors and turnover ratios could not be estimated in pre-flexion larvae, because the larvae did not attain a steady state with their diet. Moreover, isotopic signatures at early larval stages were connected to maternal transmission, whose isotopic nitrogen and carbon showed significant linear correlations with growth in length (SL) of the reared ABFT larvae until 15 DAH (Fig. 4).

By solving this relationship, it was feasible to estimate the amount of maternally transmitted stable isotopes of nitrogen and carbon, taking into account the distribution of larval sizes (SL) of pre-flexion larvae. The estimated isotopic composition of maternal origin did not differ significantly from the values of stable isotopes recorded in the egg stage (Kruskal-Wallis, p>0.05). Based on the proposed model, the maternally transmitted isotopic signatures (δ¹⁵N, δ¹³C) can be predicted from the pre-flexion (SL) larvae of ABFT (Table 1).

At 17 DAH, ABFT larvae started feeding on gilthead seabream yolk-sac larvae. This larval prey had high nitrogen isotopic values (δ¹⁵N 12.91±0.13‰) and low carbon isotopic values (δ¹³C –20.90±0.32‰). Therefore, the isotopic values of post-flexion ABFT larvae showed a significant increase in nitrogen and a decrease in carbon (p<0.05, Kruskal-Wallis). The final phase of the experiment (24-32 DAH) showed isotopic values in a similar range to the initial values recorded at the early stages (δ¹⁵N 13.30±0.35‰ and δ¹³C –18.01±0.18‰, respectively). Between 24 and 32 DAH, ABFT larvae reached the isotopic steady state of nitrogen with their diet. The discrimination factor estimated during this period was 0.4‰. By contrast, δ¹³C values did not attain the isotopic steady state with the diet (Fig. 5).

The δ¹⁵N values of the post-flexion larvae changed rapidly when their diet switched to gilthead seabream yolk-sac larvae. These values increased until the asymptotic value of the prey (12.94±0.13‰) was reached, which occurred 9 days after the start of feeding on yolk-sac larvae (Fig. 6). The estimated nitrogen turnover rate (half-time, t₅₀) was 2.49±0.28 days and the constant m was negative in post-flexion larvae (–0.11). The exponential model showed a good fit for DW, evidencing fast body mass growth (k= 0.38).

From the ecological point of view, larval growth rates are paramount to survival, as postulated by the growth mortality hypothesis (Houde et al. 1987, Anderson 1988). Larval survival is directly related to recruitment and may therefore have important fisheries management implications. The faster growth in the ABFT larvae may be largely responsible for the most successful year class (2003) during the past years (García et al. 2013, Suzuki et al. 2013). The relationship of ABFT larval growth and survival was also evidenced in Pacific ABFT larvae and juveniles (Tanaka et al. 2006, Satoh et al. 2013).

Studies of trophic ecology in early life stages of fish highlight the existence of a "critical period” in the transition from yolk-sac to exogenous feeding as one of the most important factors affecting larval survival (Fortier and Leggett 1985). Growth in this time interval is slow, and larval mortalities are high under rearing conditions (Kaji 2003). Fast growth occurs after notochord flexion, increasing significantly during the last phase of experimental study (15 to 32 DAH) (Figs 1 and 2). Overall, those larvae with 32 DAH reached an average of 45 mm SL and 265 mg DW, slightly lower than the values reported by Yúfera et al. (2014), but higher than those reported by Miyashita et al. (2001) for the Pacific BFT (Thunnus orientalis). The growth observed from reared larvae is consistent with the estimated growth of field-captured ABFT (García et al. 2013), despite the fact that the field larvae were conserved in liquid N-preserved specimens, whose shrinkage was not taken into account. However, this conservation agent (liquid nitrogen conserved larvae and sea water) accounted for the least shrinkage in sardine larvae (6% of total SL) (García et al. 2004). Thus, due to the more robust morphological larval characteristics of ABFT larvae, the age-length key of larval tuna from the field (García et al. 2013) does not differ significantly from that of reared tuna. In the field, ABFT size at post-flexion was at 5.5 mm, and considering that Itoh et al. (2000) observed the non-deposition of daily increments for a period of 4 DAH, both growth models are in agreement.

The supply of different diets or preys to reared fishes can have important consequences on growth variability (Gamboa-Delgado et al. 2008, Tanaka et al. 2010, 2014). Under rearing conditions, the trophic shift from planktivory to piscivory is crucial for larval growth (Reglero et al. 2014). At the end of the flexion stage, ABFT larvae need to predate on larger and faster preys (Ortega et al. 2011). In this experiment, the piscivorous diet began at 17 DAH and the survival after this phase was 1.5% at 21 DAH. At this stage, the maternal isotopic signatures are not evidenced because the isotopic composition of the prey is incorporated in the larval tissues (see Fig. 3). It is during this critical period of trophic diet change that the first signs of gastric glands appear, initiating the formation of the digestive system (Kaji 2003, Yúfera et al. 2014). Consequently, the growth hormone increases together with the increase in somatic mass (Kaji et al. 1999, Kaji 2003).

Tanaka et al. (2014) carried out an isotopic study that analysed two different larval diets using stable isotopes as nutritional tracers from the flexion stage to early juveniles in Pacific BFT (Thunnus orientalis). The feeding regime consisted of a group fed with rotifers and another group fed with yolk-sac larvae of the spangled emperor fish (Lethrinus nebulosus) at 15 DAH. The fish-fed group showed significantly higher growth than the rotifer-fed group. Therefore, the δ¹⁵N values increased exponentially with DAH while those of the rotifer-fed group remained at low δ¹⁵N levels, demonstrating that nutritional quality can enhance larval growth and, ultimately, larval survival.

The maternal isotopic ratios observed in the initial stages of larval development open the possibility of estimating the nutritional tracers of mature females from field-collected eggs and from the initial stages of larval development. Maternal qualities related to condition may be relevant to growth potential of larvae (Marteinsdottir and Steinarsson 1998, Green and McCormick 2005, Swain and Nayak 2009, Perez and Fuiman 2015). Under natural conditions, higher values of δ¹⁵N have shown greater growth potential in bullet tuna larvae (Auxis rochei) and in the Mediterranean anchovy (Engraulis encrasicolus) (Laiz-Carrión et al. 2011, Quintanilla et al. 2015). Thus, it is reasonable to assume that isotopic values of mother-offspring provide relevant information on the trophic position of the broodstock and may reflect larval growth characteristics as well as migratory patterns between populations (Hoffman et al. 2011, Le Bourg et al. 2014).

It is at this initial stage of larval development when the isotopic signatures of maternal transmission predominate. At this stage, larval mortality is high because food intake does not fully compensate for the somatic growth rates that lead to the gradual decline of the isotopic ratios. The role of maternal influence is preponderant for bearing healthy offspring where quality seed depends on the state of maternal nutrition (Swain and Nayak 2009, Perez and Fuiman 2015). The maternal nutritional condition showed important links with the development of survival skills developed by reared larvae of red drum (Scienops ocellatus) (Perez and Fuiman 2015). Furthermore, progeny of certain species, such as, those belonging to the family of salmonids can inherit maternal isotopic signatures depending on the spawning habitat in which these were spawned (Jardine et al. 2008).

In regard to isotopic studies, little or no isotopic discrimination between maternal somatic tissue and yolk-sac tissue has been shown (Le Bourg et al. 2014, Hoffman et al. 2011). In aplacental deep-sea sharks, small differences between mothers and their embryos were observed in the values of δ¹⁵N and δ¹³C, proving that shark embryos reflect the same trophic position as their mothers (McMeans et al. 2009, Vaudo et al. 2010).

The signatures of stable isotope of nitrogen in wild adult ABFT spawners from the Mediterranean reported in scientific literature are in the order of those estimated by egg and pre-flexion larvae (10.89±0.95‰ for wild-caught tuna and 11.83±0.28‰ for cage-reared tuna) (Varela et al. 2011). However, Estrada et al. (2005) found high values of δ¹⁵N, which reached 14‰ in adult ABFT from the Gulf of Maine. In the present experimental study, fertilized ABFT eggs collected from tuna fattening cages showed δ¹⁵N values of 12.35±0.04‰. Being able to estimate the maternal isotopic signatures from early life larval stages of tuna is an asset that cannot be disregarded. The maternal isotopic signatures may be related to larval growth and condition variability, as well as to the origin of maternal feeding grounds.

Field SIA studies on the trophic ecology on wild ABFT larvae are scarce (Laiz-Carrión et al. 2014, 2015). To understand early life trophodynamics more comprehensively, experimental research is needed. To interpret field data, field-based estimates of stable isotopes require experimental measurements of discrimination factors under laboratory-controlled conditions during the different ontogenic stages. The post-flexion larvae have the characteristics of an early juvenile. These have the necessary natatory and visual capacities to predate efficiently. Furthermore, ABFT larvae also show a precocious development of the digestive system that allows them to undergo the early diet shift to piscivory and contribute to their rapid somatic growth (Miyashita et al. 2001, Reglero et al. 2014, Yúfera et al. 2014).

Faster larval growth occurs when larvae are fed with protein-enriched sources with high δ¹⁵N values under aquaculture conditions (Gamboa-Delgado et al. 2008, Tanaka et al. 2014). In field-based isotopic experiences, faster growth in larvae of A. rochei and E. encrasicolus showing higher δ¹⁵N values has also been observed (Laiz-Carrión et al. 2013, Quintanilla et al. 2015). Higher δ¹⁵N values in field post-flexion ABFT larvae indicated a greater feeding selectivity (Laiz-Carrión et al. 2015). Our results showed much higher δ¹⁵N values in reared post-flexion larvae. These values do not reflect those observed in natural populations, where fast growth rates of captive fish can potentially result in the decrease in the discrimination factor (Δ¹⁵N) (Gamboa-Delgado et al. 2008, Varela et al. 2011).

The estimated value of the nitrogen isotopic discrimination factor (0.4‰) was lower than that reported by Varela et al. (2011) in ABFT juveniles (1.64‰). The isotopic discrimination factor is lower in the first stages of development due to the high growth rates, which can cause changes in the δ¹³C and δ¹⁵N values needed for tissue growth (Gamboa-Delgado et al. 2008). However, the isotopic discrimination factor of Δ¹³C could not be calculated because this stable isotope did not reach isotopic steady state with its prey. Therefore, we can assume that carbon requires more time to be integrated in larval tissues, as observed in other fish species (Gamboa-Delgado et al. 2008, Kiyoko et al. 2008).

Short tissue half-times are common for nitrogen in early life stages of fish (2.8-5.2 d) (Vander Zanden and Rasmussen 2001). Tissues of fast-growing animals exhibit shorter half-times (t₅₀) for nitrogen than slow-growing animals (MacAvoy et al. 2005), due to the fast growth rates characteristic of early life stages. The nitrogen isotopic changes observed in larvae can be attributed to tissue accretion and not to tissue metabolic turnover, contrary to what has been observed in adult organisms (Martínez del Río et al. 2009, Le Vay and Gamboa-Delgado 2011).

Since the carbon isotope did not reach the steady state between the larvae and its diet, it was not possible to estimate the turnover rates for carbon.

The results of this study examining the effects of diet composition, developmental stage, growth rates on stable isotope signatures represent the first data set that can be used to estimate the discrimination factors of the larval developmental stages of ABFT. These results will therefore be of broad relevance for future trophic field studies facilitating to the interpretation stable isotope data from field-captured larval specimens. In addition, the possibility of estimating maternal isotopic signatures from eggs and pre-flexion larvae opens research expectations which may relate maternal trophic levels with early larval growth potential.