Tracking zebrafish for reproducible studies

A study can be technically sound yet impossible to assess when readers do not know the zebrafish line, animal origin, developmental stage, husbandry conditions or experimental unit. The “Zebrafishology” review in Communications Biology converts these common gaps into a checklist tailored to Danio rerio. It complements ARRIVE 2.0 by connecting general animal-research reporting principles with the genetic, developmental and operational features of this model. For aquatic facilities and research teams, the work begins before manuscript preparation: useful information must be defined, measured and retained from study design onwards.

A model-specific checklist, not a regulatory standard

The authors of “Zebrafishology” offer study-design and reporting guidance. It is neither legislation nor evidence that one identical list suits every experiment. The review brings together key characteristics of the zebrafish model and turns them into questions that authors, reviewers and editors should be able to answer.

ARRIVE 2.0 supplies the cross-species framework. Its Essential 10 address study design, sample size, randomisation, blinding, outcome measures, statistical methods, animals and procedures. The Recommended Set adds context on housing, husbandry, monitoring, ethics and data access. “Zebrafishology” explains what those categories mean when dozens of larvae may come from one clutch, laboratory “wild-type” lines are genetically different and developmental stage can change rapidly with age and temperature.

This combination prevents two mistakes. The first is treating a checklist as paperwork completed at the end. The second is assuming that a large number of fish automatically makes a study robust. Reporting cannot repair an inadequate design. It allows readers to understand the design that was used, assess its limits and attempt an appropriate replication.

Identify the fish and their history precisely

“Wild type” does not describe a single genetic background. AB, TU, TLF and their substrains can differ substantially, while the background of a transgenic line may change through breeding. A usable Methods section should therefore report the standard line name, source, background of transgenes or mutations, crosses performed and, where relevant, a link to a resource such as ZFIN. An internal colony label alone is difficult to interpret outside the institution.

Age in days post-fertilisation does not always define developmental stage adequately. Developmental rate is affected by temperature and can be altered by experimental manipulation. The review recommends pairing time with appropriate morphological criteria. In adults, age, length and maturity provide complementary information. Sex or sex distribution should be stated when adults are used, with a rationale where only one sex is included.

Family origin matters as well. One hundred embryos from a single pair are not one hundred independent biological replicates. A manuscript should state the number of breeding pairs, clutches and collection days, how embryos were pooled, and how they were allocated to conditions. Without this information, readers cannot separate a treatment effect from a family or clutch effect.

Treat husbandry as part of the experiment

Conditions before testing can influence the measurements. At minimum, teams should document the housing system, stocking density, diet and feeding schedule, photoperiod, relevant water parameters, unusual changes and health-monitoring context. For embryos and larvae maintained off-system, volume, density per vessel, medium, renewal schedule, temperature and timing of manipulation become components of the protocol.

Not every variable needs the same detail in every paper. Behavioural work should make noise, vibration, illumination, test time, acclimation and tracking equipment accessible. Developmental experiments need morphological stage, density, temperature and exposure timing. Mutant studies need genotyping methods, reference assembly, relevant sequences and appropriate genetic controls.

The professional approach is to start with the scientific question and identify factors capable of changing the primary outcome. A weekly mean pH cannot replace a measurement at a critical time if a brief fluctuation may affect the endpoint. Conversely, collecting large volumes of data without a hypothesis or interpretation rule burdens the record without necessarily improving transparency.

Separate experimental units from individual fish

The number of animals, experimental units and repeated experiments are not interchangeable. In a plate, several larvae may share a well, treatment or clutch. In a facility, several tanks may share a water system. The checklist asks researchers to define what independently receives the intervention and what provides an independent observation.

Useful reporting distinguishes technical replicates, which capture measurement variation, from biological replicates, which sample variation between animals, families, tanks or days. It also explains how observations from multiple fish were combined. Reporting “n” without saying whether it means larvae, wells, clutches or repeated experiments leaves a major ambiguity.

Randomisation should be described as an operation, not merely named. Among free-swimming larvae, collecting the easiest individuals first may create selection bias. Distributing offspring from several pairs across groups, balancing plate positions and repeating the experiment on separate days are design decisions rather than writing details. Blinding, exclusions, missing data and outcome selection should likewise be decided in advance and reported.

Build traceability before data collection

Teams can make the guidance operational through a shared study record linking researchers, aquatic-facility staff and technology platforms. Each group or line should connect to a stable identifier, parents or clutch, source tank, health status, water parameters and husbandry events. The experimental plan adds allocation, experimental unit, interventions, planned exclusions, timing and responsible staff.

Instruments and software deserve comparable precision: relevant model or configuration, software version, acquisition settings, analysis thresholds, transformations and scripts. For a behavioural assay, the apparatus, geometry, acclimation period, light programme and definition of movement should be reconstructable. Raw data and analysis code can then be deposited in an appropriate repository with metadata linking every file to the experimental plan.

This organisation provides value before publication. It exposes missing fields, inconsistent identifiers and confusion about replication while the protocol can still be corrected. It also helps teams investigate drift, husbandry incidents and unexpected findings without retrospectively turning coincidence into explanation.

Limits and points of caution

“Zebrafishology” is a guidance review, not a trial comparing laboratories that did or did not use the checklist. It draws on published evidence and author expertise to recommend design and reporting practices. Some suggestions, including numbers of crosses or repeats, must be adapted to the question, data type, expected variation and statistical plan.

Transparency does not mean imposing one water recipe, one line or one protocol everywhere. Controlled differences may be scientifically useful. They need to be described so readers can see which population and conditions the conclusions apply to. Reproducibility also does not require every laboratory to obtain an identical number; it requires a result that can be understood, tested and compared within an explicit framework.

The review is broad and includes examples from genetics, development, imaging and behaviour. It cannot replace discipline-specific validation of an assay or local veterinary and ethical oversight. A reporting field should be retained because it supports interpretation, animal welfare or traceability, not simply because it appears on a generic form.

What facilities can change now

Before the next study, teams can compare their project form with the ARRIVE 2.0 Essential 10 and the zebrafish-specific checklist. Three priorities stand out: identify animals and their origins unambiguously, define the experimental unit before calculating numbers, and record the husbandry and test conditions that could influence the primary outcome. The Methods section then becomes the output of an active traceability system rather than a late reconstruction.

Vetofish can support research facilities by reviewing aquatic protocols, structuring husbandry and study records, identifying potential confounders and designing proportionate monitoring plans. The aim is to make each decision auditable without turning a checklist into a universal protocol or promising to eliminate all biological variation.