Aquariums: standardise behavioural monitoring

Watching an animal does not automatically create comparable evidence. Two teams may use the same label for different behaviours, observe at different times of day or identify individuals with unequal confidence. A 2026 paper on ZooMonitor Community describes how zoos and aquariums can build a shared framework before pooling observations. For public aquariums, the central issue is not which software to select. It is how to make husbandry decisions traceable while preserving the biological and operational differences between species and facilities.

The common seadragon (Phyllopteryx taeniolatus) project provides a particularly relevant aquatic example. Five aquariums coordinated eight months of courtship observations and environmental context. The paper presents the project design, not the biological results. It therefore offers a practical lesson in standardisation without supporting claims about optimal reproductive conditions.

Start with a shared question, not a shared app

Wark, Pham and Andrews describe ZooMonitor Community as a collaborative extension of a behavioural monitoring system. The platform can distribute projects, protocols, training resources and data across institutions. It also supports participant management and access controls. These functions address a familiar problem: locally valuable observations are often stored in formats that cannot be compared easily with records from another organisation.

Scientific design still comes first. Teams need a specific question, such as whether animals use space differently after a habitat change, whether feeding schedules alter anticipatory behaviour, or which conditions coincide with courtship. Without a defined question, a large dataset may remain a heterogeneous inventory rather than evidence for a decision.

The protocol should specify the observation unit, session length, time-of-day coverage, sampling schedule and contextual variables. Behaviour categories must be visible, operationally defined and linked to a possible interpretation. A digital platform records those decisions; it does not make them on behalf of the animal-care team.

Five aquariums followed seadragons for eight months

The common seadragon project was initiated because successful breeding in managed care has remained sporadic. US and international aquariums accredited through AZA or EAZA participated to include varied housing and husbandry conditions. Working with the species population coordinator, five aquariums developed a standardised courtship ethogram, a reference-video catalogue, observer-training materials, reliability assessments and a shared protocol.

Observers recorded how often courtship behaviours occurred, where they occurred within exhibits and which environmental and social conditions accompanied them. Data collection continued for eight consecutive months. That duration can capture patterns missed by isolated sessions, but it also creates operational demands: staff continuity, training for replacement observers, consistent session coverage and documentation of changes to exhibits or group composition.

Standardisation did not mean pretending that every aquarium was identical. When groups contained more than five seadragons, individual identification became increasingly difficult. Facilities could record the number of animals involved rather than assign uncertain identities. They could also add non-courtship behaviours needed for local welfare or husbandry monitoring. The comparable core remained fixed while local additions were documented separately.

Test observer agreement before comparing facilities

Behaviour should be defined by visible actions rather than presumed intention. “Moves towards another animal” is easier to reproduce than “seeks a mate”. Reference videos help teams agree on category boundaries, but agreement still needs to be measured.

The projects described in the paper use staged training that can include a behaviour-identification quiz, video reliability tests and in-person reliability checks. The purpose is not to create a perfect score. It is to show that observers code the same events similarly enough for the intended analysis and to expose categories that remain ambiguous. Poor agreement should lead to clearer definitions, more training or consolidation of categories that cannot be separated consistently.

Reliability is not a one-off credential. Staff change, experience develops and local habits emerge. Visibility also varies with lighting, stocking density, obstructions and observer position. A cross-aquarium comparison becomes misleading if one facility follows clearly identifiable individuals at close range while another observes a partly hidden group without recording that limitation.

Context is essential to interpretation

A difference in behavioural frequency is not automatically a difference in welfare. Time of day, temperature, lighting, maintenance activity, group structure, reproductive phase and observer visibility can all affect what is recorded. Relevant contextual variables should therefore be selected before data collection and captured at a level of precision that teams can sustain.

Behavioural records do not replace veterinary examinations, feeding data, body condition or water-quality measurements. They add temporal and spatial information about when a change occurs. A single signal calls for verification. A repeated pattern that aligns with other indicators may justify a review of the exhibit, husbandry schedule or clinical plan.

Data governance also belongs in the protocol. Teams should know who can inspect raw sessions, correct a record, export a dataset or identify an institution. ZooMonitor Community documentation states that sharing is not automatic, participating organisations retain ownership and projects can request identifiable or pseudonymised contributions. Facilities should assess those choices before joining a project, especially where internal operational data are involved.

What the 2026 paper does—and does not—show

The publication is a communication describing a platform and several collaborative projects. For common seadragons, it documents the observation framework but does not yet report which environmental factors were associated with courtship or reproductive success. It cannot support a universal temperature target, preferred social composition or superior husbandry practice.

Five aquariums provide more variation than a single-site project, but they do not represent every facility. Small populations, local husbandry choices and unequal visibility remain important constraints. Coordination software cannot remove sampling bias, confounding or the need for appropriate statistical analysis.

The paper does show that a shared framework can accommodate real operational limits. Two participating aquariums that had not previously used ZooMonitor adopted it for continued monitoring of their seadragons and other species. That is evidence of method uptake, not proof that clinical welfare or reproduction improved.

Six decisions for an actionable monitoring programme

Before multi-institutional recording begins, partners should agree on six elements: one professional question; an explicit observation unit; a shared core ethogram; comparable temporal coverage; an observer-reliability threshold; and an interpretation rule connecting behaviour with other indicators. A short pilot using common videos and then live sessions in each exhibit can reveal unusable categories before months of work accumulate.

The protocol should also anticipate deviations. An unidentifiable animal, relocated camera, exhibit change or technical interruption should not be silently repaired in the dataset. It should be logged as context or trigger exclusion under a predefined rule. Transparent missingness is more valuable than an artificial impression of completeness.

Finally, each indicator needs a proportionate response. A change might trigger additional observations, a veterinary review, a water-quality check or a structured discussion among keepers. It should not automatically generate a diagnosis. Multi-institutional collaboration is most useful when it helps distinguish a local anomaly from a broader pattern worthy of further study.

Turning observation into a traceable decision

Shared behavioural monitoring can move public aquariums beyond isolated impressions, provided they standardise before comparing. The common seadragon project demonstrates a workable compromise: a common core, trained observers, contextual data and documented flexibility when individual identification is unreliable.

Vetofish can help aquarium teams define the professional question, develop an ethogram, select complementary welfare indicators, train observers and interpret behavioural, veterinary and environmental information together. This also requires clear reporting and decision pathways for aquarium teams. The aim is not to maximise the number of observations. It is to produce evidence robust enough to inform a decision and cautious enough to preserve uncertainty.