The operational welfare indicators for tilapia were organized according to four of the five freedoms for animals established by the Farm Animal Council (24): environmental, health, nutritional, and behavioral. The fifth freedom, psychological, is intrinsically evaluated through the behavioral and physical analysis of the animals. The methodology employed followed the same principles already used in the creation of the protocols previously developed by our group for O. niloticus (23) and grass carp, Ctenopharyngodon idella (25), during the grow-out phase of both and also for white-leg shrimp, Penaeus vannamei throughout its entire production cycle (26).

An integrative systematic review (27) was conducted using the Google Scholar platform as the research base. The aspects related to the environment, health, nutrition, and behavior associated with the species, as well as the specific welfare indicators and their respective reference values for each cultivation stage (Figure 1), were studied and defined. For the grow-out phase of tilapia, emphasis was given to cultivations carried out in earth ponds, the primary fish farming system used globally (18). The research included books, technical and scientific articles, case studies, manuals, and technical reports developed by international institutions, as well as theses and dissertations. Materials containing the following terms were selected: “Oreochromis niloticus” AND welfare indicator AND production stage, in the titles, abstracts, or keywords. The welfare indicators defined for each stage of the cultivation process are listed in Table 1. The search period extended from 1985 to 2023.

The reference values for each indicator were classified through a system involving three possible scores (1, 2, or 3). A score of 1 indicates the limits of variation of a particular indicator within the ranges considered ideal for the target species. A score of 2 pertains to variations the animals tolerate, which can cause deleterious effects, provided they are non-lethal. A score of 3 indicates significant levels of variation in a specific indicator that significantly compromises the health and even the survival of the animals, which is deemed unacceptable from an animal welfare perspective. The maximum tolerated mortality rates, the primary indicator of the degree of welfare of the fish, were set at levels much lower than those found in nature, taking into account each stage of the production cycle. The reference values for each indicator and score were established based on the literature available for the species.

An experimental trial was conducted to establish health protocols for the early life stages of tilapia and to test the pre-selected indicators, allowing for the assessment of their applicability under operational conditions and the evaluation techniques for each indicator. The experiment was conducted at the Laboratory of Research with Aquatic Organisms (LAPOA) of the Integrated Group of Aquaculture and Environmental Studies (GIA) at the Federal University of Paraná, Curitiba, Brazil. All husbandry and experimental procedures were approved by the Animal Use Ethics Committee of the Agricultural Sciences Campus (CEUA) of the Federal University of Paraná (protocol number 021/2023).

A total of 480 newly hatched tilapia larvae were used, subdivided into 12 aquariums of 30 liters in volume, each linked to a chemical–biological filtration system, under controlled conditions of temperature (27.01 ± 1.0°C), pH (7.83 ± 0.2), and dissolved oxygen (5.40 ± 0.4 mg/L). Over 15 days, a tilapia larva from each aquarium was randomly selected using a catch net and carefully and individually transferred to cell culture dishes, duly labeled, containing 10 mL of water on a daily basis. This procedure ensured that the samplings were representative, giving greater precision to the results. The collected fish were anesthetized with clove oil at a concentration of 500 mg/L and kept until they reached stage V anesthesia (~5 min). At that point, the fish have a medullary collapse and permanent unconsciousness (28). The cell culture dishes containing the collected animals were then transferred to the imaging laboratory, where the animals underwent physical evaluation procedures and photographic recording to determine their health status and evaluate the welfare indicators. The following organs were assessed under a stereo microscope (Figure 2): eyes, mouth and jaws, skin, fins, gill covers, spinal column, and yolk sac. Next, an individual and bilateral photographic record of each individual was made. The indicators that proved unfeasible to measure under operational conditions and on a commercial scale were excluded from the final version of the protocol. After this final verification, the protocols were reorganized in the format and content presented in Tables 1–10.

Calculation of tilapia welfare indices utilized the same mathematical equations proposed by Pedrazzani et al. (25) for evaluating the welfare degree of C. idella. The weights assigned to each indicator in the respective indices were established based on the number of valid bibliographic references found for O. niloticus in each production phase, using Google Scholar as the research platform. The variable “Y” was calculated as the integer part of the natural logarithm (ln) of the number of articles identified through specific keywords, as shown in Equation 1. Consequently, partial welfare indices (PWIs) were proposed for each category, along with the general welfare index (GWI), calculated from the PWIs. The PWIs were computed using Equation 2, which considers the weights assigned to each indicator and their respective scores.

where:

PWI: Partial welfare index standardized to vary continuously between 0 (critical risk of harm to farmed fish welfare) and 1 (maximum welfare or, otherwise, minimum risk of injury to animal welfare), regardless of the number of indicators used in each freedom.

X: Freedom (En, environmental; Be, behavioral; Nu, nutritional, or He, health).

Y: Weight assigned to the specific indicator.

S: Score assigned to the indicators in the analyzed fish farm.

The general welfare index (GWI) was calculated as the arithmetic mean of the PWIs, multiplied by an elimination factor (kl, Equation 3). The kl is defined based on the observed mortality rate. Thus, mortality becomes mathematically the most critical indicator for measuring the welfare degree of farmed fish in captivity. By definition, if the mortality rate exceeds 30%, the value of the elimination factor will be equal to zero (kl = 0), automatically indicating a “critical” classification for the GWI of the evaluated fish. If the mortality rate is below 30%, the eliminatio n factor will be equal to 1 (kl=1), and the welfare of the fish will be determined based on the respective indicators analyzed, their scores, and weights.

where:

WGI: General welfare index, que varia de 0 (critical risk of harm to farmed fish welfare) and 1 (maximum welfare or, otherwise, minimum risk of injury to animal welfare).

kl: Knockout level (risk of total impairment of the degree of welfare).

The partial confidence levels (CLx) proposed for each PWI are determined based on the number of indicators effectively analyzed in the field. The more indicators evaluated compared to the proposed indicators, the higher the confidence level of the results. The general confidence level (GCL) is calculated as the arithmetic mean of the CLx (Equation 5). Finally, the PWIx, GWI, CLX, and GCL are classified and interpreted according to the obtained values.

CL_X: PWIx confidence level.

∑W_An: Sum of the weights of the indicators analyzed for the freedom x.

∑W_max: Sum of the weights of all the defined indicators for the freedom x.

In the scope of environmental freedom, a set of 12 indicators has been established for different cultivation phases (Table 2). The physicochemical indicators of water have been standardized across all stages. This allows for the prevention of shocks during the transfer of animals between phases whilst maintaining strict adherence to the adopted environmental indicator reference values. Predators and aquatic and terrestrial cohabitants have also been included as environmental welfare indicators, with scores assigned based on their control or presence/absence. A score of 1 should be considered when there is no evidence of other terrestrial or aquatic species in the fishpond. In the case of terrestrial predators, a score of 2 should be applied when the fish farmer adopts control measures, such as filters or screens, but indirect contact still occurs (e.g., a visual connection between tilapia and their predators). For aquatic interspecific predators or cohabitants, a score of 2 should be applied to polyculture systems. The evaluator should assign a score of 3 when there is no control over predators or the presence of cohabitants from other species. The photoperiod should be considered a relevant indicator during the reproduction and larviculture phases. In the reproduction phase, additional indicators such as hapa cleanliness and the proportion of males and females used in tanks and fishponds were considered (Table 2).

Four relevant nutritional indicators applicable to all cultivation phases have been defined: crude protein content in the feed provided to the fish, feed quantity with the biomass of the batch, feeding frequency, and feeding distribution range in the respective cultivation system. The reference values for these indicators were determined based on the weight of tilapia (Table 5). Considering the physiological and energy demands and the management practices adopted during the reproduction phase, two sub-stages were established for calculating the amount of feed provided to the broodstock: “maintenance” and “breeding”. In the grow-out phase, the feed conversion rate (FCR) was incorporated in addition to the mentioned indicators.

The weights assigned to determine the importance of each adopted welfare indicator were established based on the number of publications identified in the Google Scholar platform. These values were obtained through a combination of general search terms (“Oreochromis niloticus” AND “aquaculture” AND the respective life phase) and specific search terms presented in Tables 8–10.

Figure 3 illustrates the application of the welfare protocol for O. niloticus during the grow-out phase. The data were simulated using a Microsoft Excel spreadsheet and are derived from a hypothetical but commonly observed scenario in commercial tilapia farming. Based on the example, partial welfare indices related to the environment (PWI_En), health (PWI_He), nutrition (PWI_Nu), and behavior (PWI_Be) of the fish are calculated. In the analyzed, the first three indices indicate a moderate level of welfare for the cultivated fish. However, the behavioral index suggests a low level of welfare due to improper management practices during harvesting and slaughter. The simulated scenario calculated the general welfare index (GWI) at 0.59, considered moderate. The confidence level in this result was the highest, as all indicators were analyzed. Examples of these index calculations during the breeding and nursery phases are included in Supplementary Figures 1, 2.

When we originally proposed the method and indices that our group has already applied to assess the degree of welfare in grass carp and white-leg shrimp (25, 26), we emphasized that the indicators, their reference values, scores, and weights would need to be periodically reviewed and updated as scientific knowledge advances on the subject. In this article, we put this concept into practice by applying our metrics to assess the welfare of tilapia not only during the grow-out phase but throughout all life stages, whilst also revising and advancing the knowledge generated previously. The indicators now applied to tilapia in the grow-out phase had already been tested and validated by our group under field conditions (23). In this article, in addition to revising and updating the indicators and reference values for this cultivation phase, we simplified the evaluation structure by reducing the number of scores for each indicator from four to three. This approach improves field evaluation, making it more objective and dynamic than protocols with higher scores (87, 140). This reduction in scores regarding changes affecting eyes, gills, and fins, for example, which often indicate pain and significant diseases, reduces subjectivity in interpreting moderate lesions, thus enhancing evaluation accuracy. However, the blood glucose indicator was removed from the current protocol due to the invasiveness of the method. Additionally, it should be considered that the blood sampling procedure itself can alter the parameters, causing acute stress to the animals.

Regarding environmental indicators such as temperature, pH, and dissolved oxygen, it is necessary to consider that these parameters naturally vary throughout the day in cultivation ponds. However, the animals can adapt (141) as long as the changes occur within tolerable limits for the species. To avoid conflicts between different values considered acceptable in the literature, we adjusted the new scores to reflect these possible changes in water quality in tilapia cultivation ponds.

The stocking density indicator was removed from the current protocol. Although it is evident that stocking density directly influences the degree of welfare and fish health (142–144), establishing fixed values for this parameter is highly subjective. Defining the ideal density regarding welfare depends significantly on the characteristics of the fish, environmental and nutritional resources provided to them, management practices, fish size, genetic characteristics, and other factors (145).

We also found several operational and conceptual difficulties regarding the shading indicator proposed in the previous protocol. Recent studies present conflicting data that do not accurately reflect the reality of pond cultures, mainly due to inadequate consideration of light intensity and its impact on water quality (148, 149). Initially, we proposed a uniform percentage of shading over the surface of the net pen or pond, typically achieved by using protective and shading screens. However, ponds often have localized shade caused by trees or topographic features in their surrounding areas. This led to misinterpretations during the application of the protocol since localized shading is detrimental to the welfare of tilapia (150). Therefore, we chose to exclude this indicator from the current protocol.

In the original protocol for tilapia grow-out, we established crude protein content (CP), feed conversion ratio (FCR), condition factor (K), and feeding behavior as nutritional indicators. However, after reviewing the data from applying the protocol in various countries (Pedrazzani, unpublished data), we found significant genetic variability amongst cultivated O. niloticus strains, which led to morphological variations in the fish. Sometimes the strains were naturally broader than long, and vice versa, which affected the value of the condition factor (K) without any relation to fish welfare. Therefore, adjusting the formula for each population or strain of the same species proved impractical for standardization.

The nutritional freedom indicator has now been adjusted to calculate welfare indices by category. Thus, to facilitate tracking the feeding history on farms, we standardized four indicators for all cultivation phases: feed quantity, protein content, feeding frequency, and feed distribution within the pond. Due to the difficulty of capturing and weighing fish in early cultivation, FCR monitoring was suggested only for the grow-out phase.

The behavioral indicators during feeding, harvesting, and slaughter were kept the same as in the original protocol, but their scores were readjusted, aiming to reduce subjectivity during the evaluation, as previously discussed regarding animal suffering associated with health freedom.

An essential conceptual aspect concerns fish slaughter, which we associate with the grow-out phase. Although the industrial-scale slaughter of tilapia is already a reality (151, 152), globally, this practice remains an exception rather than the rule. Transporting water and live tilapia to the processing plants is costly and requires complex logistics and appropriate slaughter facilities. In most cases, fish are sold alive in markets or slaughtered on-site at the farms where they are cultivated (29, 39, 57, 152–154). Under these circumstances, it is common for animals to be slaughtered without prior stunning, often asphyxiated in the air or ice (155–157). Thus, we believe that a more practical way to improve fish welfare is to consider slaughter as part of the grow-out process. However, considering the expansion of international tilapia trade, the evolution of industrial aquaculture, and the increasing interest of consumers in the quality of the product and the welfare of farmed fish, it is plausible to project that slaughter will soon become a genuinely autonomous stage in the tilapia production cycle.

The sanitary indices employed during the grow-out phase were maintained with those developed for adult fish and further expanded. The additions were intended to integrate essential management practices commonly used in commercial tilapia reproduction and larviculture facilities. Indicators such as the stage of sexual maturation, the time interval adopted for the recovery/conditioning of breeders between reproductive cycles, and the assessment of the adoption or not of methods or mechanisms of control to avoid inbreeding amongst breeder batches were included. In addition to these, other less intuitive indicators were proposed, such as, for instance, the maximum intervals adopted for cleaning the hapas where the breeders are kept during reproduction—an essential factor to prevent the obstruction of the screens since this hinders the renewal of water and compromises the health of the breeders and can cause not only a reduction of zootechnical indices but also a decrease in immunity and the emergence of diseases (30, 57, 158). The proportion between males and females used during reproduction was another indicator included in the protocol since it interferes with population dynamics and the degree of aggressiveness of the males during the mating phase (30).

For the behavioral assessment of the breeders, we emphasize the recommendation for using anesthesia during any invasive procedures. We kept the “feeding behavior” indicator, although it is relevant to note that female tilapias reduce their food intake during the mating and spawning periods by incubating the eggs in the mouth, whilst males can increase their feed intake in the same period (159). This protocol included the photoperiod due to its central role in the natural induction of sexual maturation and in defining reproductive rates (160, 161).

Concerning the welfare of eggs, larvae, and post-larvae, considering that this is still a controversial topic, that it is in its initial stages of unveiling scientific knowledge, and that there is a tremendous natural vulnerability of larvae and post-larvae to handling, which requires even greater caution and rigor; we advocate for the implementation of the protocol proposed in this phase of tilapia cultivation. However, at the same time, we suggest that the suitability and necessity of its application be determined on a case-by-case basis, considering the overall conditions of the batch and the local structural and operational capacity to assess the proposed indicators.

The welfare of eggs and larvae involves the parents' nutritional, social, and environmental experiences during their development (162). The environment likely influences the epigenetic pattern of gametes, embryos, or adult organisms (163). During gametogenesis, the DNA is reprogrammed, and this information will be transmitted to the offspring, resulting in transgenerational effects that directly impact the quantity, viability, social status, neurogenesis, and adaptation of future generations (162). Sneddon et al. (1) highlight that fish larvae have various brain structures that process emotions and learning, although they are not identical to the human brain. Lopez et al. (164) demonstrated that zebrafish larvae at 5 days post-fertilization (5 daf) respond to harmful and potentially painful stimulation caused by environmental acidification, exhibiting similar behaviors to adult fish and reducing their activities. This response was alleviated by analgesic drugs such as lidocaine and morphine. Furthermore, different larval rearing protocols can have a significant impact on larval size and mass, survival rates, and the sex ratio of larvae (165). Therefore, it is necessary to consider that welfare should be understood as continuous and intergenerational, as there is a direct link between offspring adaptation and the resources provided by parents (162, 166). On the other hand, the quality of the eggs and larvae will also significantly impact the welfare and health of tilapias throughout their lives (167).

Another argument to be considered is that, despite the legislation and regulatory frameworks of the vast majority of countries still not protecting fish larvae, we must consider the scientific arguments linked to the presence of sentience in larvae and fry (1, 162, 164). In this sense, we advocate the application of the proposed protocol based on the precautionary principle, which establishes that when evidence of sentience is inconclusive, we should “give the benefit of the doubt” to the animal or “err on the side of caution” (168). Thus, we understand that the theme “welfare in the early stages of fish life” has relevance to be applied in current aquaculture but, mainly, that it will have a significant impact on the aquaculture that will be practiced in the coming years, possibly under a scenario of regulatory restrictions and rigorous governance practices (169). Therefore, be it for biological, ethical, moral, or commercial reasons, and even recognizing the fragility of the current stage of knowledge on the subject and the need for subsequent discussions on the effectiveness of the application of welfare protocols for the larval and post-larval stages of O. niloticus, we understand it to be recommendable and, at the same time, almost “inevitable” that the early stages of life be included in animal welfare assessment protocols in fish farming.

We tested and validated their operational feasibility in the laboratory to assess the indicators proposed here for the early stages of tilapia life. The experiments carried out made it possible to identify the most suitable indicators and, at the same time, exclude those that did not meet the established prerequisites.

Some health indicators were not incorporated into the larval protocols, as structures such as fins and gills are still in development during the ontogenetic processes that occur during the nursery phase, making their visual assessment difficult. For post-larvae and fry, it is relevant to consider the “degree of emaciation” as an indirect indicator of feeding effectiveness. The “yolk sac” indicator was included in the assessment of larvae, as organisms at this stage of life still have endogenous energy reserves and, therefore, do not show apparent signs of emaciation (76). In all life stages, we kept the mortality rate as an indicator of welfare, as most fish deaths in captivity are likely preceded or accompanied by suffering. Thus, long-term mortality rates may serve as indicators of the degree of retrospective welfare and signal possible future impacts on the success rates of the enterprise (170).

In the analysis of tilapia larvae and post-larvae, we identified a significant number of articles focussed solely on swimming behavior and the changes commonly related to water contamination or the occurrence of diseases. Therefore, we included only this behavioral indicator for the larval stage once they have endogenous feeding. We also excluded some indicators from this stage, as larvae are produced in laboratories, rendering indicators such as the presence of “terrestrial predators” or “water transparency” irrelevant, for instance.

Animal welfare should not be linked to cultural differences or subjective criteria but to the species' biology (171). Therefore, quantitative animal welfare assessment is essential for promoting humane and responsible management practices in the animal production industry (172, 173). Moreover, using quantitative and standardized approaches in welfare measurement allows producers to tangibly demonstrate their commitment to animal welfare, which helps build consumer trust and generate new market opportunities (174, 175).

The metrics proposed in this article aim, pioneeringly, to provide a holistic and quantifiable assessment of the welfare of Nile tilapia throughout all stages of its captive life cycle. These metrics were established based on indicators that were simultaneously simple, understandable, and already part of the routine production of the species on a commercial scale. To achieve this, we used indicators representing the nutritional, behavioral, health, and critical environmental conditions to which tilapia are exposed throughout their production process.

Furthermore, this study's partial (PWI) and overall (GWI) welfare indices offer an objective animal welfare assessment. They are based on data and scientifically supported metrics rather than opinions or subjective factors, providing excellent reliability and accuracy. The proposed indices can provide producers with a valuable tool for retrospective and prospective analyses within the same production cycle, enabling informed strategic decisions for the welfare of farmed fish and the profitability and efficiency of their businesses.

There is already recognition within the scientific community that a single score simplifies data interpretation and constitutes a valuable tool for researchers, producers, certifying bodies, and regulatory agencies (87, 122). This characteristic allows the proposed indices to establish a solid foundation for developing animal welfare regulations and guidelines, enabling authorities to define clear and measurable standards to ensure ethical and humane treatment in tilapia farming operations, one of the most critical species in global aquaculture (18).

In this study, PWIs and GWIs follow the same conceptual and mathematical logic applied and extensively discussed concerning grass carp (25) and white-leg shrimp (26). However, the indicators and their respective reference values, scores, and weights are specific to O. niloticus, covering aspects of breeding, larviculture, fingerling rearing, and the grow-out phase—in this case, in earthen ponds. Like in previous studies, the weights assigned to each indicator were identified using Google Scholar. These weights ranged from 2 to 10 and were determined based on the number of scientific documents related to each indicator. It should be noted, however, that despite being practical, this method has limitations. For example, the number of publications on a specific topic may not reflect its relevance in a practical context; it may underestimate the importance of less-researched welfare indicators; it can be influenced by the availability of funding for research in specific areas, which may not reflect the importance of those areas for animal welfare; there may also be variations due to the language used in search terms and differences in the consulted databases. Despite these limitations, the approach is robust, standardisable, and encompasses several advantages, including the comprehensiveness and objectivity of evaluating welfare in tilapia farming, recognizing differences that each indicator presents in fish welfare, and the ability to update and refine as new research is published. Subsequent analysis can explore alternative methods for assigning weights to welfare indicators and further examine how the relative importance of these indicators may vary across different life stages.