The effect of laboratory diet and feeding on growth parameters in zebrafish

Despite being one of the most used laboratory species in biomedical, behavioural and physiological research, the nutritional requirements of zebrafish (Danio rerio) are poorly understood, and no standardised laboratory diet exists. Diet and feeding regimen can significantly impact the welfare of the fish and in turn experimental reproducibility. Consequently, the establishment of a standardised diet and feeding protocol for laboratory zebrafish is imperative to enhance animal welfare, guarantee research reproducibility and advance the economic and environmental sustainability of laboratory dietary practices. The aim of this systematic review was to determine the optimal feed for juvenile zebrafish growth and development. A comprehensive search was conducted in PubMed, Scopus and Google Scholar to identify relevant studies published up to August 2023 and the studies were selected based on the predefined inclusion/exclusion criteria. A total of 1065 articles were identified in the databases, of which 14 were included in this review. We conducted data extraction and risk-of-bias analysis in the included studies. Statistical comparisons for specific growth rate, weight gain (%) and length gain (%) parameters were performed to determine the optimal feed for enhanced juvenile growth. We identify an insect-based diet as optimal for juvenile growth for all three growth parameters. We also identify areas of potential heterogeneity and conclude by encouraging a standardised laboratory diet to ensure reproducible data and encourage zebrafish welfare.


Introduction
Zebrafish (Danio rerio) are one of the most common research organisms in the fields of behaviour, genetics, physiology and biomedical science with high genetic homology to mammals, low husbandry costs, rapid breeding and an ease of genetic manipulation [1][2][3] .
However, despite their increasing use in research, there is a lack of standardised feeding regimen and knowledge on nutritional requirements [4][5][6][7] .The lack of consistent diet used between research groups can significantly affect fish welfare and increase experimental inconsistencies 4,8 .It is therefore essential that a standardised diet is established which promotes welfare of the zebrafish with the added benefit of improving environmental and economic sustainability of feeding.A lack of comprehension regarding the requirements of the optimal zebrafish diet has limited the ability to determine a standardised laboratory diet.Currently, the dietary requirements of laboratory zebrafish are related to published information on other species of fish that demonstrate similar feeding habits or live in a comparable habitat to wild zebrafish 4,[9][10][11] .Fig. 1 provides a basic overview of the known dietary requirements of zebrafish.It is widely accepted that dietary protein is the principal constituent for growth and development of zebrafish due to its deposition as lean body mass tissue which contributes to overall weight gain.The lipid requirements of zebrafish are unknown; however, it is assumed that they require similar amounts to other fish species.On average current diets for zebrafish constitute anywhere between 10 -18% lipid content of dry matter 4,12 .Carbohydrates are a principal energy source for zebrafish and therefore are considered an essential constituent of the diet.
In addition, zebrafish require a range of vitamins and minerals for a balanced diet including calcium, sodium, iron, riboflavin and biotin, although the exact requirements remain to be elucidated 4 .Like other species, any deficiencies in vitamins and minerals can result in metabolic and structural abnormalities.Although it is agreed that zebrafish require these dietary constituents for a balanced and effective diet, the overall diet composition remains unknown.
Therefore, it is essential to determine a standardised laboratory diet through establishing the nutritional requirements of zebrafish to ensure reproducible research and promote welfare.In this review we aimed to establish the optimal feed for three different parameters of growth (specific growth rate (SGR), length gain and weight gain) in juvenile zebrafish.To determine this, we performed a systematic review of the available scientific literature studying diet on growth in juvenile zebrafish.We provide an overview of the current literature relating to zebrafish growth responses with different feeds by providing a qualitative description of the published studies as well as evaluating the impact of bias arising from methodological conduct, reporting quality and selective publication.In summary, our aim was to provide a comprehensive and data-driven approach to investigate the following research question: What feed-type is optimal for growth in juvenile zebrafish?

Methods
This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines 13 .All data and our analyses are accessible and downloadable on the Open Science Framework (OSF) (https://osf.io/d5f7t/?view_only=8445745e4716467a9997bc62700ee67f).

Search Strategy
Searches were carried out in three bibliographic databases --PubMed, Scopus and Google Scholar --using keywords which relate to our research topic for the intervention (fish feeds) and the desired population (laboratory zebrafish).Therefore, the following search string was applied: 'Zebrafish AND ('feed' OR 'diet' OR 'feeding' OR 'food') AND ('growth' OR 'survival' OR 'reproduction' OR 'welfare').The search was carried out with no limitations on language or start date until August 2023.The reference lists of the included studies were also screened to detect additional relevant articles.

Eligibility screening
After searching the databases, the selection of studies included in this review was performed by one independent researcher (C.H.) with regular discussions with a second reviewer (M.P.).Titles/abstracts were initially screened to identify and exclude duplicates.Thereafter, studies were selected based on the inclusion and exclusion criteria (see below) by reading the titles/abstracts and, where necessary, the full text.Several studies had restricted access (paywalls), and the corresponding authors were contacted by email to request copies of the paper.They were given 14-days to provide the full-text of the article, after which the study was excluded from the analysis.
Studies were included if they met the following criteria: (1) original experimental research performed with adult zebrafish ≥ 28 dpf testing different fish feeds or feeding regimens; (2) studies reporting growth (SGR, length and/or weight) and/or reproductive effects of fish feeds; (3) articles were peer-reviewed; (4) fish ≥ 28 dpf but ≤ 4 months old at the start of the feeding trial (growth analysis); (5) fish > 4 months at the start of the feeding trial (reproductive analysis).Exclusion criteria were: (1) use of zebrafish < 28 dpf or other research organisms; (2) studies which incorporated over-or under-feeding; (3) assessment of food supplements for translation into higher organisms; (4) review articles, retracted articles, book chapters, scientific letters, and conference abstracts.
We required a minimum of five studies for both growth and reproduction analyses.However, during this stage we discovered that there were too few reproduction-based studies which fit our inclusion criteria and therefore the reproductive analysis was removed (See Section 3.1.Search Results for further information).

Data extraction
One investigator (C.H.) worked independently to complete the data extraction, and consultation was provided, when necessary, by an additional reviewer (M.P.).All data extraction was performed from the full text and figures and, where required, data were recalculated into the desired format (i.e., percentage length gain was calculated from initial and final length data).The following information was collected: (1) general data: title, authors, publication year, age of fish, strain of fish, sex split in the experiment, sample size per experimental group (n), tank density, diet and feeding regimen prior to experimental diet, experimental diet and feeding regimen and husbandry conditions (See Table 2); (2) growth results where applicable including: standard growth rate (SGR), percentage weight gain, percentage length gain and percentage survival rate (see Supplementary Data Table 4).Data were extracted as mean ± SE and when papers reported standard deviation (SD), this was converted into SE using an equation previously described in literature and used in previous meta-analyses 14 .Some studies only presented data in graphs (or not at all); therefore, authors were contacted via email to provide the additional data required for our analysis.These authors were given a 14-day period to respond, thereafter, PlotDigitizer (version 2.6.9.www.polotdigitizer.com)was used to manually estimate numbers from the graphs.Coauthorship networks were constructed using VOSviewer software version 1.6.20 (www.vosviewer.com) 15.

Risk of Bias and Reporting Quality
To evaluate the quality of included studies, a risk-of-bias assessment was conducted by three independent investigators (C.H., A.C. and P.R.) for each paper.This analysis was performed using the SYstematic Review Centre for Laboratory animal Experimentation (SYRCLE) risk of bias tool for animal studies 16   (10) any other potential biases.The reporting was as follows: low (green), slight (light green), moderate (orange), high or unclear (red).Bias plots were created using GraphPad Prism 10 (GraphPad Prism version 10.1.2for Windows, GraphPad Software.Boston, Massachusetts USA, www.graphpad.com).

Data analysis
The studies were split into three groups (SGR, percentage weight gain and percentage length gain) depending on data availability.Many of the identified studies were designed as a 'control' feed vs 'experimental' feed(s) experiment.Because there was no consistency between the 'control' diets, we further grouped the feeds together into six distinctive categories: (1)  where appropriate, a one-way Analysis of Variance (ANOVA) or unpaired student's t-test was performed to determine the most effective feed within the category using GraphPad Prism.
For studies where more than one feed was available within a category, the feed reported as the most effective within the study was used for the analysis.A final one-way ANOVA was conducted combining the optimal feeds from each category to determine the overall optimal feed category for SGR, percentage weight gain and percentage length gain.Type 2 error rates were p**** < 0.0001, p*** < 0.001, p** < 0.01, p* < 0.05.Fig. 2 provides a detailed breakdown of this analytical process.

Fig. 2. Flow chart of analytical steps use for statistical comparisons.
Note: An overview of the steps taken to perform the analysis presented in this paper.Figure was generated using Biorender.com.

Search Results
Database searches resulted in an initial 1065 documents (n = 512 from PubMed, and n = 553 from Scopus) through systematic searches (Fig. 3).Duplicates (n = 656) were removed and studies with unrelated research questions, different research model, and other study types were excluded (n = 590).After these exclusions, 66 articles remained, which were subjected to a full-text reading.Any studies which exclusively used zebrafish < 28 dpf, incorporated supplements into the diet for translational purposes or over and/or underfed the fish for obesity studies were excluded (n = 41).After selection by title, abstract and full text, 25 original research articles with adult zebrafish ≥ 28 dpf testing the effects of feed on growth were obtained.Fourteen of these studies were included in the present systematic review and the main characteristics are found in Table 2 and additional extracted data in supplementary data Table 4.The four studies that were not included having been initially deemed suitable were excluded because they produced significant publication bias (n = 1) or were the wrong age for the parameters for which they presented data (n = 3) 8,17-19 .Four studies were identified which assessed reproduction; however, this was below our threshold for inclusion for statistical comparison and therefore were not included any further in the analysis [20][21][22][23] .

Fig. 3: Flowchart diagram of the collection of studies and selection process.
Note: The PRISMA flow diagram for the systematic review detailing the database searches the number of abstracts and full texts screened, the number of reports retrieved, and the number of studies included in the review.

Study Characteristics
The studies included in this review were published between 2012 and 2023.These studies were carried out in the United States of America (USA) (n = 3), Italy (n = 2), Turkey (n=2), India (n = 3), Brazil (n = 2), France (n = 1) and Portugal (n =1).The ages of fish ranged from 28 dpf to 4-months at the start of the feeding trial and trial length ranged from 30 to 210 days (Fig. 4).
A breakdown of the diets can be seen in Table 1.

Table 1:
The different feeds used in the articles including the feeding category and the number of groups using this feed (n).

Feed
Category n Gemma Micro 300 (GM300)  Note: Box-and-Whisker plots comparing the age at the start of the feeding trial (A) and the length of feeding trial (B).
The studies reported significant effects on juvenile growth with differing zebrafish feeds with no effect on overall survival observed [24][25][26][27] .A common finding amongst authors was a positive correlation between protein content, pro-biotic, Chlorella spirulina (Chlorella sp.) concentration and insect-based diets with body weight 24,25,28,29 .Similar findings were reported on for fish length.However, feed intake was reported to decrease with increasing protein content whilst a positive relationship was observed between body weight and lean body mass with the diets 28,30,31 .In addition to growth and survival parameters, Dhanasiri (2020) reported moderate transcriptome changes in fast-muscle samples in zebrafish fed plant-based diets compared to animal-based diets.Vural (2021) also highlighted up-regulation of growth hormone genes with Royal Jelly supplementation compared to an animal-based diet.A description of the studies included in the review can be found in Table 2.More detailed information at the study level for the variables extracted is available in Supplementary Data Table 4.The co-authorship network analysis can be found in Figure 5, showing collaboration between researchers or research groups relating to feeding effect on zebrafish growth. 32mited collaboration was identified, see the interactive version of this co-authorship analysis for further information (www.vosviewer.com) 15.

Risk of bias
The overall risk of bias for the items evaluating the methodological quality of included studies was considered low, except for items 3-7, and 10 (Fig. 6).Due to this, the overall publication risk of bias presented here is considered unclear.For items

Growth parameters
Zebrafish growth and development is greatly impacted by feeding regimen and protein levels; however, no standardised laboratory diet exists to date 7,34,35 .Therefore, here we aimed to determine the optimal feed type for juvenile larval growth by performing sub-category analyses within each feeding category (Supplementary data 4-6) followed by an overall statistical analysis to determine the optimal category (Fig. 7).Optimal growth was defined here as the largest SGR, greatest percentage weight gain and percentage length gain.Gemma Micro 300 and Artemia nauplii; GM300: Gemma Micro 300; SGR: Specific Growth Rate; 1/day: once per day; [number]: category.Data is presented as mean ± SEM.

Percentage weight gain
Sub-group analysis for category 2 -5 percentage weight gain revealed 100% Chlorella sp.
(category 2), 50% FM (category 3), defatted prepupae meal (category 4) and 6.4% Royal Jelly (category 5) to be optimal within each respective category.The final analysis also included 1 The full statistical output for all analyses can be found on the OSF for this review.
The overall analysis revealed a significant effect of diet on weight gain in juvenile laboratory zebrafish (One-way ANOVA: F(5, 424) = 32.09,p < 0.0001) (Fig. 7B).Further analysis with Tukey's multiple comparison test revealed an insect-based diet (defatted prepupae meal) to be optimal for percentage weight gain compared to all other feeding categories, closely followed by animal-based diet with 50% FM 37 .No significant difference was seen between the other feeding categories 25,27,29,36 .

Percentage length gain
Sub-group analysis revealed 50% FM (category 3) and GM300 combined with brine shrimp (category 6) to be optimal within their respective categories for juvenile length gain and were therefore compared against wheat gluten (category 2), defatted prepupae meal (category 4) and 9% FO + 0.5% CLEO (category 5) in the overall analysis 26,30,35,37 .The analysis demonstrates a significant effect of diet on juvenile length gain (One-way ANOVA: F(4, 877) = 279.6,p < 0.0001) (Fig. 7C).Further analysis with Tukey's multiple comparison test revealed an insect-based diet to be optimal for percentage length gain compared to the other feeding categories (Fig. 7C).
The post-hoc comparison also revealed each diet to differ significantly between each other except for wheat gluten compared to 9% FO and 0.5% CLEO 26,30 .

Discussion
In this review, we aimed to determine the optimal feed for laboratory juvenile zebrafish growth and development.We report here that an insect-based diet, specifically defatted prepupae meal, is optimal for juvenile growth and development with significantly increased SGR, percentage weight gain and percentage length gain compared to all other tested categories (Fig. 7A-C).We also highlight the significant effect different diets can have upon growth and development of zebrafish.This signifies the importance of determining a standardised diet to promote reproducibility and experimental consistency between zebrafish laboratories.
We incorporated three different growth parameters to ensure numerous aspects of growth were accounted for within our analysis including SGR, percentage weight gain and percentage length gain.SGR is a common growth statistic used in the zebrafish community which determines the growth (unit measurement non-specific) per day 27 .This parameter should account for the large variability in the length of feeding trial that we identified, which cannot be accounted for using weight and length gain (Fig. 4).Length and weight gain were also selected due to their commonality between the included studies.Although here we determined optimal growth to be the greatest increase in all three parameters, we do acknowledge that this may not necessarily constitute the optimal endpoint.A previous body condition scoring (BCS) has been derived for laboratory zebrafish to assess health and welfare 39 .The authors identified the optimal width of zebrafish to be between 8-10 mm 39 .Although width was not accounted for here due to lack of reporting in studies, it highlights the importance of ensuring an optimal weight for welfare and health.
The defatted prepupae meal comes from the black soldier fly (BSF) (Hermetia illucens L.), which has previously been identified as a promising replacement of FM in zebrafish diet due to its comparable amino acid profile 37 .The diet was generated from an established colony and following hatching were reared on organic waste 37 .This highlights one of the unique environmental advantages of an insect-based diet because rearing on organic waste can reduce environmental load of waste streams 40 .Furthermore, studies have identified the BSF for significantly reducing greenhouse gas emissions, which highlights its potential for having a substantially low carbon footprint 41 .Additionally, incorporation of BSF in zebrafish feed could alleviate the pressures placed on fishery resources when wild fish are farmed to feed laboratory zebrafish for FM-based diets 42 .Therefore, from an environmental perspective the incorporation of defatted prepupae meal to zebrafish diets is a suitable option.
However, the fatty acid (FA) profile of insects does not always meet the nutritional requirements of laboratory fish because insects are normally rich in saturated FA's and monounsaturated FAs, rather than polyunsaturated FAs typically required by zebrafish for a balanced diet 42 .This is a significant concern which must not be overlooked when considering the incorporation of insect-based diets for laboratory zebrafish.Insects also have chitin in their exoskeleton, which can reduce food intake, digestibility and nutrient absorption because of its limited digestive capacity 42 .This was supported by Lanes (2021) who observed increased chitinase gene expression levels in adult zebrafish compared to their control fish fed a FM based diet.They also report that chitinase gene expression was dependent upon the BSF development stage.However, our findings presented here suggest that not only is an insectbased feed suitable for juvenile zebrafish growth and development but appears to result in substantial increases in growth compared to all other diet types.Like previous studies on zebrafish fed with BSF, we observed no significant effects on zebrafish survival (see data in OSF) 42,43 .However, previous studies have identified that insect meal can reduce fish growth and welfare over long periods of time with a high percentage of dietary inclusion (> 25%) 43- 46 .This is suggested to be a result of the FA imbalance from the BSF 37 .Lanes (2021) suggests that defatted BSF meal mitigates this imbalance between the FA composition of highly unsaturated and saturated FA, which is evidenced here in this review and in their research.
Furthermore, a study tested defatted BSF up to 100% with Atlantic salmon (Salmo salar) and found comparable growth responses to control diets 47 .Therefore, defatted BSF meal could potentially provide an economical and environmental alternative to FM with improved growth performance in juvenile zebrafish.
Although our findings suggest a significant effect of including insect-based diets for juvenile zebrafish growth and development, we acknowledge caveats to this finding.
Firstly, age can have a significant effect on growth responses 48 .Although steps were taken to reduce the likelihood of age influencing findings by providing a strict age range for the inclusion criteria, this may still have impacted the findings.The average start age between the included studies was 46 dpf (Table 2 and Fig. 4).A study performed by Singleman & Holtzman (2014) suggests that zebrafish growth is greatest between the ages of 30 -45 dpf compared to 45 -60 dpf.However, the largest increase is seen between 90 -180 dpf 48 .Lanes (2021)   began their study at 30 dpf and therefore may have seen a greater increase in growth compared to the average starting age of 46 dpf.The average length of feeding trial also varied significantly with Lanes (2021) having one of the shortest (Fig. 4).In addition, Lanes (2021) did not account for any sex differences in growth (Table 2).Female zebrafish tend to grow heavier and longer than male zebrafish and therefore an unequal split in sex may have impacted the results 49,50 .However, 50% of the included studies did not disclose a sex split and one of the studies that did only used males 26 .Other factors which may have significantly impacted the growth findings include strain of fish, water quality, temperature, availability of feed and population density 48 .All these factors differed significantly between the included studies (Table 2).This high heterogeneity between the included studies means the results presented here should be interpreted with caution.Furthermore, the main effects of the analyses were influenced by studies with a high risk of bias (Fig. 6).Although efforts were made throughout to improve the reporting quality of the research, the publication of studies adhering to measures designed to mitigate the risk of bias associated with methodological conduct is still low.
Despite the high heterogeneity and potential risk of bias between the studies reported on, we believe our findings to be of importance within the zebrafish community.Therefore, we suggest that further studies should be undertaken to identify long-term and generational effects of an insect-based diet on zebrafish.This includes genetic and behavioural assays as well as the impact of feeding on fecundity and offspring longevity. 40We also encourage the standardisation of zebrafish husbandry and feeding to ensure experimental reproducibility and encourage enhanced welfare.

Conclusions
In conclusion, our systematic review highlights the importance of identifying a standardised diet and feeding regimen for juvenile laboratory zebrafish growth and development.The lack of nutritional control not only raises welfare considerations but also concerns relating to the reliability and reproducibility of research outcomes.Our statistical comparison highlighted the significant variability in growth responses among studies with different feeding diets.However, our findings suggest that an insect-based diet, particularly defatted prepupae meal sourced from the BSF, holds promise for optimal growth and development of juvenile zebrafish.Despite the environmental and economic advantages of insect-based diets, challenges such as FA imbalance and chitin content must be addressed.
While our review contributes valuable insights, caution is warranted due to substantial heterogeneity among studies and potential biases.Future research should focus on long-term effects, generational impacts, and comprehensive assessments of feeding strategies to ensure the welfare and scientific integrity of zebrafish research.Addressing these concerns is paramount for advancing the understanding of zebrafish biology and maximizing the model's potential in various research fields.

Fig. 4 :
Fig. 4: Comparison of the age at the start of the feeding trial and length of feeding trial represented using box-and-whisker plots.

Fig. 5 :
Fig. 5: Co-authorship network analysis of researchers that authored studies assessing the effect of diet on growth.

Fig. 7 .
Fig.7.The overall growth analysis for the three growth parameters.

Table 2 : Qualitative description of studies reporting growth-related effects of experimental diets on juvenile zebrafish.
Note: Description of the studies included in the present meta-analysis.Key; BW: bodyweight, CS: Chlorella spirulina, CLEO: clove leaf oil, dpf: days post-fertilisation, D(1): diet(number), FM: fishmeal, FO: flaxseed oil, FPH: fish protein hydrosylate, GM300: Gemma Micro 300, HIM: Hermetia illucens meal, L:D : Light: Dark, N/A: not applicable, PET: pet-store zebrafish, RJ: Royal Jelly, SPC: soy-protein concentrate, SPI: soyprotein isolate, WG: Wheat gluten 3, 5 and 7 100% of the studies provided insufficient data to rule out biases arising from the allocation of the animals to experimental groups, or biases resulting from investigators.Items 4 and 6 had moderate bias with 26.6% and 86.6% of the studies respectively providing insufficient information to rule out the potential of biases impacting results due to random housing of experimental animals and random outcome selection.Gonzales (2012)was found to have a high risk of bias for item 10 due to 50% of the experimental diets incorporating feeds which are no longer commercially available.Therefore, this study was removed from further analysis to ensure the results are of commercial relevance.Additionally, Samuel (2021) provided insufficient information regarding the age of fish with age defined as 'uniform sized adults' as well as a lack of husbandry condition reporting.However, this study was included in the final analysis.The eight other additional potential risks were for Barca (2023), Fronte (2021), Lanes (2021), Sevgili (2019),Fernandes (2016)and Smith (2013) who did not include a sex split and for da Silva (2020) and Karga & Mandal (2016) who did not include the strain of fish.These pose a potential risk of bias which must be taken into consideration during analysis and interpretation of results.Out of the 450 scores for risk of bias, there were 21 (4.7%) disagreements between the three independent investigators.Of the 21 disagreements, 1 (4.8%) was for item 2, 6 (28.6%) were for item 4, 5 (23.8%) were for item 6 and 9 (42.8%) were for item 10.