Pirarucu larviculture in green water provides heavier fish and modulates locomotor activity

The green water technique uses microalgae in the water of indoor larviculture, providing a darker environment to favor fish growth, welfare and health. We evaluated growth performance and locomotor activity after light exposure of pirarucu ( Arapaima gigas ) larvae reared in green or clear water. During one test, pirarucu larvae (3.6 ± 0.3 cm; 0.36 ± 0.1 g) were reared in 50-L circular tanks (n = 3 per treatment, 50 larvae per tank) in a static system containing green water [microalgae (w3algae; Bernaqua ® 10 g m -3 ) added] or clear water (control). Fish weaning was achieved by co-feeding with Artemia nauplii and microdiets for seven days until full microdiet substitution. Larvae were biometrically evaluated on days 10, 17 and 24 to assess growth performance. In a second test, the locomotor activity of the larvae was analyzed before and after light exposure (1400 ± 60 lx) for 48 h according to an ethogram. After 24 days, the larvae reared in the green water were significantly heavier than those from the clear water, and displayed significantly fewer circular swimming movements. Body cortisol increased in both groups after light exposure. The microalgae provided an additional food source for larvae, with positive impact on growth until day 17 of larviculture. Green water can be a strategy to achieve better results in pirarucu larviculture, especially during and up to 10 days after the co-feeding period.


INTRODUCTION
The larval period is the fish's life phase that requires most care because it is when the main morpho-physiological transformations occur, such as complete yolk sac depletion, mouth and anus opening, the differentiation and functionality of internal organs, body and eye pigmentation, and fin and scale appearance (Halverson 2013;Portella et al. 2014).
Microalgae have been employed as additives in feed formulations to provide highly unsaturated fatty acids, and to enrich water during larviculture, a technique known as green water, which has provided good results in marine fish larviculture (Izquierdo et al. 2006;Roy and Pal 2015).The benefits to larvae are related to the nutrients absorbed directly by microalgae ingestion, or indirectly by consumption of filtering microcrustaceans that are enriched with microalgae nutrients (Rocha et al. 2008).Microalgae (Chlorella pyrenoidosa Starr & Zeikus, 1987 and Chlorella spp.) contain vitamins A, C, D3 and E, and highly unsaturated omega-3 fatty acids like eicosapentaenoic and docosahexaenoic acids, EPA and DHA, respectively (Drewery et al. 2014).EPA and DHA are essential in fish early larval development (Derner et al. 2006) as they participate in the formation of cell membrane phospholipids, and are responsible for maintaining cell integrity, fluidity and permeability (Prieto et al. 2006).
The health condition of larvae is crucial for their development and some behavioral characteristics are indicators of animal quality (Dias et al. 2004).The environment's color is one of the factors that can increase or depress behavioral patterns (Fanta 1995;Papoutsoglou et al. 2000;Merighe et al. 2004).For example, black, white, yellow and red should be avoided in tilapia (Oreochromis niloticus Linnaeus, 1758) farming as they cause stress or significant changes in fish behavior (Merighe et al. 2004).Green, however, is similar to the color of natural environments and does not interfere with animal behavior (Merighe et al. 2004).Accordingly, we hypothesized that the green water technique would reduce the physiological response resulting from a stressor (e.g.light intensity) and increase fish growth.
In outdoor ponds, the presence of suspended inorganic (clay, silt and carbonate) and organic (plankton and small organisms) particles affects turbidity and light penetration in the underwater environment (Yi et al. 2003;Villamizar et al. 2011).Optimal light conditions during fish larviculture increase growth and survival, and promote normal development (Villamizar et al. 2011).However, under outdoor conditions, fish larvae are more susceptible to the action of predators (piscivorous birds, aquatic insects, bats) (Gonçalves et al. 2019).Dominant larvae stand out in food competition, making access to food difficult for submissive larvae, which can become smaller and eventually die (Lima et al. 2017).
In current pirarucu (Arapaima gigas Schinz, 1823) larviculture, larvae remain in outdoor ponds and are cared for by parents until they reach 7-10 cm in size, when they are collected and allocated to indoor tanks to be fed commercial diets (Pereira-Filho et al. 2010;Halverson 2013).During the period under parental care, larval mortality rates can reach around 90% (Ono et al. 2004;Pereira-Filho et al. 2010;Gonçalves et al. 2019).
An alternative to achieve higher zootechnical performance and survival rates of pirarucu larvae is the capture of larvae when they have an inflated vesicle and swim close to the breeder's head, transfer the larvae to the laboratory and train them to receive formulated feed by weaning, which can increase larval survival up to 95% (Araújo da Silva et al. 2018;Gonçalves et al. 2019).Weaning consists of gradual feed transition, also known as co-feeding, and is characterized by progressively reducing live food concomitantly with increasing inert feed supply until the fish have adapted to exclusive inert food ingestion (Azevedo et al. 2016).The co-feeding strategy is used not only to promote the intake of formulated feed, but also to stimulate the development of the digestive system (Engrola et al. 2009).
Rearing fish in the laboratory allows greater management control and easier observation of behavioral indicators of health, stress and hunger (Johansen et al. 2006).For example, swimming speed is inversely proportional to the amount of food present in the rearing environment, so that the observation of the swimming behavior can be related to the food satiety level (Nunn et al. 2011).We evaluated zootechnical performance, locomotor activity and response to stress factors of pirarucu larvae reared in green water and in conventional environment (clear water system) during the weaning period.

MATERIAL AND METHODS
This study was approved by the ethics committee on animal experimentation and research of Instituto Nacional de Pesquisas da Amazônia (INPA), Manaus, Amazonas, Brazil (protocol # 016/2016 CEUA/INPA).The experiments were carried out at the INPA's aquaculture experimental station at the Coordination of Technology and Innovation (COTEI).The pirarucu larvae were obtained from natural spawning in a fish pond at Santo Antônio farm (AM-240 highway, km 48, Amazonas, Brazil), and were collected when their gas bladder inflated and they began to swim next to the breeder's head and immeadiately transported to INPA.
Two experiments were performed, referred as Test I and Test II.In Test I, we evaluated the zootechnical performance of larvae reared in green or clear water during the weaning period.In Test II, the behavior and body cortisol of the Test-I larvae were evaluated in relation to light exposure as a stress factor.

Test I
The test was carried out in a completely randomized design with two environments: clear water (CW) (no microalgae were added) and green water (GW), with the inclusion of microalgae (w3algae; Bernaqua®), with Chlorella pyrenoidosa and Chlorella spp. at a concentration of 10 g m -3 , with three replicates for each treatment.For each replicate, pirarucu larvae (3.6 ± 0.32 cm; 0.36 ± 0.06 g) were transferred to circular polyethylene tanks with a useful volume of 50 L and 50 larvae per tank, in a static system with 50% water cleaning and renewal twice daily, during 24 days.The stocking density followed Santana et al. (2020).The water quality parameters observed during the 24-day experiment were: temperature = 26.52 ± 2.57 ºC; pH = 6.50 ± 0.28; dissolved oxygen = 6.46 ± 0.31 mg L -1 ; nitrite = 0.29 ± 0.15 mg L -1 ; carbon dioxide = 15.66 ± 8.77 mg L -1 ; ammonia = 2.28 ± 0.40 mg L -1 .All the parameters fell within the comfort range for pirarucu (Chu Koo et al. 2017).Regarding ammonia, pirarucu tolerates high levels of ammonia, as in a similar study in a static system, pirarucu juveniles tolerated an ammonia concentration of 25 mg L -1 (Cavero et al. 2004).The ammonia levels in our study did not interfere with larval growth.
During the first seven days, weaning was performed by co-feeding with Artemia nauplii (3000 nauplii larvae day - 1 ) and microdiet pellets (Sparos®) with 200-400 µm size, which were offered until day 10.From day 11 to day 24, larvae were fed microdiet pellets of 400-600 µm.Both pellet types contained 63.51% crude protein, 15.34% fat, 1.70% crude fiber and 8.10% ash, according to the manufacturer's information.Larvae were fed 10 times a day, and the daily feeding rate was planned according to the daily growth rate and feed conversion in previous trials.

Test II
After Test I ended, the larvae (6.70 ± 0.49 cm; 1.75 ± 0.47 g) were placed into aquaria (to make filming possible) maintaining treatment and replicate identity.Each aquarium had a useful volume of 8 L (six aquaria with CW and six aquaria with GW larvae; eight larvae per aquarium).Larvae remained in the same treatments as in Test 1 in a 2x2 factorial scheme in two rearing environments (CW and GW), with and without light exposure (three aquaria per treatment).The same management procedures as in the previous test were applied.The experiment began after a 48-hour period for larvae to acclimatize to their new environment.Three aquaria of each water treatment (CW and GW) were exposed to a light intensity of 1400 ± 60 lx for 48 h, which is considered a stressor for fish larvae (Lopes et al. 2018).The other three aquaria of each water treatment were not exposed to the light.Larvae behavior was recorded (Intelbras ® camera VHD 1010 B G3) for a 1-hour period before and after light exposure.The locomotor activity of three larvae from each tank was quantified for a 10-minute period (Lopes et al. 2018) (from minutes 30 to 40) by quantifying the frequency of movements (rotational, vertical, horizontal), static moments (larva stopped) and random contacts (Olla et al. 1978;Sabate et al. 2008).Total movements corresponded to the sum of the frequencies of all movements (rotational, vertical and horizontal), described in an ethogram applied to matrinxã (Brycon amazonicus Agassiz, 1829) (Souza et al. 2014) and adapted to pirarucu larvae (Table 1).

Behavioral unit Description
Circular swimming movement (CSM) Fish move from one place to another in a circle Immediately after the end of Test II, three larvae from each aquarium were euthanized by physical methods (CONCEA 2018) and macerated individually in a porcelain crucible), dissolved in diethyl ether, centrifuged (3500 rpm, 5 min), dried in nitrogen vapor and stored at -20 ºC.For body cortisol reading procedure, samples (total weight after freezing: 0.9 -1.2 g) were suspended in 1 mL of PBS and plate assembly followed the manufacturer's directions.The reading was done by ELISA (Cortisol ELISA kit -DRG Diagnostics), which has been tested and validated for fish by Santamaría and Casallas (2007) and by Canavello et al. (2011).

Statistical analysis
All response variables of both tests had normal distribution (Shapiro-Wilk test) and variance homogeneity (Levene test).The zootechnical performance variables (on days 10, 17 and 24) were submitted to a one-way ANOVA (N = 3 tanks per ACTA AMAZONICA treatment).The frequency distribution of size classes was compared between treatments with a Chi-square test.
Behavioral variables and cortisol were compared among treatments with a two-way ANOVA, considering each larva as an observation unit (nine larvae per treatment, three larvae per tank).When interaction between factors (water type and light exposure) occurred, a pairwise comparison was performed using the Tukey test .When the interaction were non-significant, the factors were evaluated individually.The significance level was 5% in all analyses.

RESULTS
After 10 and 17 trial days, the larvae reared in GW had significantly higher weight and length than CW larvae (Figure 1).On day 24, however, no statistical difference was observed for total length and growth performance, but GW larvae (1.97 ± 0.10 g) remained significantly heavier than CW larvae (1.79 ± 0.16 g) (Figure 1 and Table 2).Rearing water did not interfere with the frequency distribution of larvae in size classes, with over 75% of GW and CW larvae classified as medium size after 24 days (Figure 2).
In Test II, there was an interaction effect between water type and light exposure for vertical movement (F = 4.4053; df = 32; p = 0.044), horizontal movement (F = 8.2228; df = 32; p = 0.007) and total movements (F = 12.4940; df = 32; p = 0.001).CW larvae showed significantly higher movement rates than GW larvae when exposed to light.
Larvae exposed to light showed more circular swimming movement (F = 4.9079; df = 32; p =0.034) and less static   moment than those not exposed to light (F = 48.933;df = 32; p <0.0001).Circular swimming movements and contact by chance were significantly higher in CW compared to GW (Table 3).The body cortisol level was significantly higher with light exposure independently of the rearing water (Figure 3).

DISCUSSION
The higher final weight of GW larvae was probably related to the intake of vitamins and fatty acids provided by the microalgae, unlike the CW larvae that did not receive this nutritional support.Pirarucu larvae can filter microalgae through gill traces, which they use as a feed source (Ono et al. 2004) until the juvenile phase when the weight is around 300 g (Lima et al. 2018).The application of lyophilized or inoculated microalgae to marine fish larviculture environment helps to stabilize water quality (Navarro and Sarasquete 1998), to feed larvae, and to maintain both the nutritional value of live food and larval development during weaning (Ferreira 2009).
After weaning (on day 10), the total length of larvae had increased by almost 49% in GW and by more than 45% in CW.Fish larvae performance depends on the quantity and nutritional quality of live food (Portella et al. 2012).Pirarucu larvae showed a similar length increase of 47.9% CSM: circular swimming movement; VM: vertical movement; HM: horizontal movement; TM: total movements; SM: static moment; CC: chance contact; L: light exposure; W: rearing water.Means followed by different lower case letters indicate significant differences for this behavior between larvae exposed or not to light.Different upper case letters in columns for each behavior and light treatment indicate significant differences between water treatments (two-way ANOVA and Tukey test; p ≤ 0.05) ACTA AMAZONICA after being fed Artemia nauplii for 15 days (Araújo da Silva et al. 2018), and 93% after being fed Artemia nauplii and zooplankton for 11 days (Alcântara et al. 2018).Pirarucu larvae fed only Ostracoda-rich zooplankton displayed an increased length of 62.7%, but lower survival (40%) than larvae fed zooplankton rich in cladocerans, copepods and rotifers (Gonçalves et al. 2019).The zooplankton from natural environments (cladocera, copepoda, rotifera) is the main food item during the larval and juvenile periods of carnivorous fish like pirarucu (Lima et al. 2018).Abundant microcrustaceans with predominance of Cladocera were found in stomach contents of pirarucu juveniles weighing up to 500g reared in earth ponds (Lima et al. 2018).Wild freshwater zooplankton ia an excellent nutritional source for fish larvae and juveniles, but its abundance is very much dependent on climatic conditions, it is potential vector of diseases, and no adequate technology is available to allow its low-cost mass production in freshwater (Vega-Orellana et al. 2006).Artemia sp. in nauplii stage offer good nutritional value and can be produced on a large scale within 24 hours, and their cysts are easily accessible on the market (Samat et al. 2020).However, Artemia nauplii represent most of the production cost in fish larviculture (Jomori et al. 2005), which is why the feed transition by co-feeding makes economic sense, in addition to the benefits of stimulating inert food intake, digestive system development, and improving growth and survival (Engrola et al. 2009;Portella et al. 2012).
The type of larviculture water did not affect larvae survival rates, which were around 76-78%, but were lower than those observed in other studies on pirarucu larviculture (93 -99%) (Alcântara et al. 2018;Araújo da Silva et al. 2018;Gonçalves et al. 2019).This could be due to the initial weaning period and the live food type supplied before the feed transition period, as pirarucu larvae fed Artemia nauplii and zooplankton until day 11 showed 99% survival after 21 larviculture days (Alcântara et al. 2018) Survival in the initial pirarucu hatchery is related to many factors such as available food, size at the beginning of weaning, density in the tank, sanitary and feeding management (Gonçalves et al. 2019).The use of microalgae in the larviculture rearing environment has been reported to shorten the feed-transition period without altering production rates, allowing to achieve higher survival rates, lower cost per larva and higher production yields than the traditional larviculture system (Jomori et al. 2005).We observed no difference in larvae survival between treatments, but the weight and length of larvae after feed transition (at 11 and 17 days) were higher in GW than in CW.In Paralichthys dentatus Linnaeus, 1766, no difference in growth performance of larvae reared in GW and CWwas observed after 42 days, but survival was higher in GW (76.1%) than in CW (27.8%) (Bengtson et al. 1999).
Pirarucu larvae are commercialized by length, and fish farmers are more interested in larvae that have already been trained to consume commercial feed.After weaning, fish are more morphologically and physiologically developed and, consequently, are more resistant (Rebelatto Junior et al. 2015;Lima et al. 2017).On day 17 of larviculture, the average total length of our larvae reared in GW was 6 cm, and they were already being fed only commercial feed, which eases management and reduces feed costs and production time.
Pirarucu larvae form schools that move in synchrony to feed (Harvelson 2013), and larvae housed in circular tanks show circular swimming movements, which are fast when fish are hungry, and slow down after feeding, indicating satiety (the authors, pers.obs.).Thus the lower swimming frequency in circular movements in GW in our study likely indicates that larvae were less hungry due to the presence of microalgae as a food source, which is further evidenced by the higher weight of the GW larvae.In addition, greater locomotor activity of CW larvae implies more energy expenditure (Gerry and Ellerby 2014), which may also have contributed to the lower weight of larvae in this treatment.GW larvae moved less and presented less chance encounters with each other than CW larvae, which potentially reduces injury rates due to interaction among larvae, which serves as a gateway to pathogens that can increase mortality rates (Huntingford et al. 2006).

CONCLUSIONS
Our results indicate that the inclusion of microalgae in water provides an additional food source for pirarucu larvae, with positive impact on larva growth until day 17 of indoor larviculture.The green water technique can be a strategy to achieve better results in pirarucu larviculture, especially during and until 10 days after the co-feeding period.

Figure 1 .
Figure 1.Evolution of the total length (A) and total weight (B) of pirarucu (Arapaima gigas) larvae reared during 24 days in clear water (CW) and green water (GW).Points indicate the mean and bars the standard deviation of three replicates (50-L tanks with initial population of 50 larvae).Letters at each time-point indicate whether the means differed significantly according to an ANOVA F-test.

Figure 2 .Figure 3 .
Figure 2. Frequency distribution of pirarucu (Arapaima gigas) larvae in size classes after 24 days reared in clear (CW) and green water (GW).Columns are the mean and bars the standard deviation of three replicates.

Table 3 .
Number of behavior units of pirarucu (Arapaima gigas) larvae reared in clear (CW) and green water (GW) with or without light exposure.Values are the mean ± standard deviation of nine replicates.