Introduction

Shrimp farming is increasingly important worldwide1,2, with shrimps among the most common marine products consumed in the European Union3. Among cultured shrimp species, Litopenaeus vannamei is the most successful due to its ease of hatchery production, fast growth, delicious flesh, high nutritional value and disease resistance4,5. According to recent reports, 6.8 million tonnes were produced in 20221,6, with an estimated 440 billion shrimp standing stock annually. Asian nations dominate global production, although imports from these producers have occasionally been associated with poor labor conditions, lack of hygiene, and use of illicit products1,7,8,9,10. Beyond these factors, the increasing social awareness is also demanding for reduced carbon footprint, aiming for local production at the European level11.

To cope with non-tropical climate conditions, indoor recirculation systems have been designed and built. These systems are energy-intensive, but prevent deforestation, wastewater discharge, disease outbreaks, soil erosion, and predation, with the advantage of supporting high stocking densities12,13. Shrimps from land-based production sites are sold as high-quality products in Europe, reared under strict regulations, ensuring food safety, sustainability and traceability. The proximity of production sites to the consumer market enables rapid distribution of fresh product.

Despite recent technological advances, crustacean welfare is often disregarded and a lack of recognition persists within the sector. Debates about the level of sentience of this subphylum have recently arisen14,15. Mellor and Reid16 developed the concept of good welfare, which illustrates the five freedom domains in the field of nutrition, physical environment, health, behavior, and mental needs. Following a detailed report from Briffa et al.17, Great Britain has already declared decapods as sentient animals and some countries, such as Germany and Switzerland, have included decapods in their Animal Welfare Acts. An expansion of existing regulations to a global scale may occur soon. As part of the animal welfare topic, the need to measure animal health and condition in husbandry routine at commercial scale is rising, through the identification of species-specific operational welfare indicators18.

Welfare impairments are caused by inadequate conditions, which induce short or long-term stress to animals. In principle, stressors trigger a cascade of reactions assigned to three response levels. The primary stress response is composed of neuroendocrine factors, which are secreted to reconstitute homeostasis. Among these hormones, crustacean hyperglycaemic hormone (CHH) is the most relevant, mobilizing energy needed for adaptation19,20. During the secondary stress response, metabolic changes occur that shift energy to restore homeostasis. These changes involve shifts in hemolymph glucose, lactate, total proteins, accelerated respiration and heart rate, as well as changes in hemocytes21,22,23. When stress persists beyond adaptive function, significant impairments at the organismal level can occur, such as immune system dysfunction and oxidative stress, transitioning chronic stress into distress. Like other invertebrates, crustaceans lack an adaptive immune system and rely solely on innate immunity24,25. Two major effector systems involved in immune activation are the phenoloxidase system, responsible for melanin or quinone synthesis, and antimicrobial peptides such as lysozyme26,27. Similarly, akirin is a protein that acts as a nuclear factor, by activating antimicrobial peptides and antioxidant genes28,29. Oxidative stress arises from an excessive accumulation of reactive oxygen species (ROS), leading to an imbalance in oxidative metabolism and subsequent cellular damage. The antioxidant defense mechanisms against ROS include enzymatic systems such as superoxide dismutase (SOD) and glutathione peroxidase (GPx), both of which play a crucial role in detoxifying superoxide radicals30,31. As an antioxidant response regulator, p53 is a protein that acts as a transcription factor involved in regulating cell processes, such as DNA repair32. Additional stress biomarkers include heat shock proteins (HSPs), a highly conserved and stress-sensitive group of proteins, and transforming growth factor-beta (TGF-β), a family of anti-inflammatory cytokines. Both HSPs and TGF-β are synthesized in response to various stressors, contributing to cellular protection and immune modulation33,34.

The tertiary stress response affects the performance of the animals, from growth rate to survival, as well as the behavior and biological impairments. Survival rate and growth are addressed by multiple studies concerning chronic stress, in the field of nitrate exposure, pH and temperature variation and resulting in reduced survival rate and/or slower growth and productivity35,36,37. Studies to date on behavioral patterns of this species have focused on feeding38,39,40. Indeed, abnormal behaviors such as swimming abnormalities, high levels of cannibalism, and escaping behavior are mentioned but are not subject of rigorous quantitative analysis in literature41. In the field of shrimp morphology, a variety of lesions are described and biological impairments such as abnormal rostrum, uropods have been associated with fungicide exposure, damaged antennae, exoskeleton and rostrum with stocking densities41,42,43.

Stocking density is a major factor influencing the growth and survival of farmed shrimp. While high-density shrimp grow-out systems offer advantages in terms of yield, they are generally associated with reduced growth and lower survival rates44,45. High stocking densities are hypothesized to disrupt homeostasis due to increased social interactions and physical injuries, leading to stress and increased susceptibility to disease46,47,48. However, the underlying physiological mechanisms remain poorly characterized, with existing studies inconsistent on alterations in biochemical and immunological parameters15,44,49,50. Moreover, the impact of high stocking density on overall health status and behavior remains largely unexplored in shrimps.

Compensatory growth is defined as the accelerated growth phase following a period of reduced development due to feed deprivation or stress. It has been documented in L. vannamei under chemical stressors such as nitrate exposure and temperature fluctuations35,50, but no data currently exist regarding compensatory growth in response to stocking density stress. Beyond growth performance, the plasticity of physiological markers, such as wound healing and behavioral recovery, requires further investigation to elucidate their role in resilience to stocking density stress. Thus, the aim of this study is to investigate various stress responses in L. vannamei exposed to different stocking densities, and take a holistic approach to assessing its effects on growth performance, welfare, and physiology.

Materials and methods

Experimental design and system management

The recirculation aquaculture system (RAS) used in the experiment consisted of twelve 600 L tanks and a water treatment unit with a biofilter, UV disinfection, protein skimmer, drum filter and ozone generator. The flow rate in the tanks was about 600 L/h, and the water renewal in the system was approximately 2% d−1. Shrimps of about 10 g were stocked in the experimental units at a density of 130 shrimp/m3 for an acclimation period of three weeks. After this period, the densities were adjusted one day before the start of the experiment, and three experimental groups were tested, with four replicates each. Animal ethics approval and welfare measures allowances were released by The Senator for Health, Women and Consumer Protection (Free Hanseatic City of Bremen), with the file number TA22-01.

The experiment was divided into two phases of 21 days each. During the first phase (stress) shrimp (mean weight 14.73 ± 0.36 g) were reared at three different stocking densities: 1 kg/m2 (low), 2 kg/m2, (standard) and 4 kg/m2 (high), corresponding to 1.83 kg/m3, 3.67 kg/m3 and 7.33 kg/m3, respectively. On day 21, the first sampling (see 2.5) was performed. The day after, shrimp were counted and weighed, and the stocking density in all experimental units was adjusted to 0.8 kg/m2 (1.33 kg/m3), and the second phase (recovery) began. After 21 days, the final sampling was carried out.

Oxygen saturation, temperature, pH, and salinity were monitored daily. Ammonia (NH3-N mg/L), nitrite (NO2-N mg/L), and nitrate (NO3-N mg/L) were determined with a QuAAtro39 Continuous Segmented Flow Analyzer (SEAL Analytical GmbH, Norderstedt, Germany) twice a week. Shrimps were fed with a commercial diet (ALLER AQUA 2.0 mm pellet) three times a day at 2.5% of body weight, adjusted twice a week according to counted mortalities and feed consumption. Before the initial, stress and recovery phase sampling, shrimp were starved for 24 h.

Biometric indices and growth performance

The survival rate, specific growth rate, weekly growth rate, relative weight gain, hepatosomatic index (HSI), and condition factor (K)51 were calculated using the following equations:

$${\text{Survival}}\;{\text{rate}}\left( \% \right) = {1}00 \times \left( {{\text{Final}}\;{\text{number}}\;{\text{of}}\;{\text{shrimps/Initial}}\;{\text{number}}\;{\text{of}}\;{\text{shrimp}}} \right)$$
$$\text{Specific growth rate (SGR) (\%/day)} = 100 \times \frac{\ln(\text{Final mean body weight}) - \ln(\text{Initial mean body weight})}{\text{Time (days)}}$$
$$\begin{aligned} & {\text{Weekly}}\;{\text{weight}}\;{\text{gain}}\left( {{\text{WWG}}} \right)\left( {{\text{g}}/{\text{week}}} \right) \\ & \quad = \left( {{\text{Final}}\;{\text{whole}}\;{\text{weight}} - {\text{Initial}}\;{\text{whole}}\;{\text{weight}}} \right)/{\text{Time}}\left( {{\text{week}}} \right) \\ \end{aligned}$$
$$\begin{aligned} & {\text{Relative}}\;{\text{weight}}\;{\text{gain}}\left( {{\text{RWG}}} \right) = \left( {{\text{Final}}\;{\text{whole}}\;{\text{weight}}} \right. \\ & \quad - \left. {{\text{Initial}}\;{\text{whole}}\;{\text{weight}}} \right)/{\text{Initial}}\;{\text{whole}}\;{\text{weight}} \\ \end{aligned}$$
$${\text{HSI}}:{\text{hepatopancreas}}\;{\text{weight}}/{\text{total}}\;{\text{weight}}$$
$${\text{K}} = {\text{Weight}}/{\text{Length}}^{{3}} \left( {\text{in cm}} \right) \times {1}00$$

Morphological observations

A review of morphological alterations was prepared based on the most characteristic modifications reported41,42,43. An injury score index was applied to nine randomly selected shrimps per tank, according to the shrimp injuries observed (Table 1). Representative examples are provided in Supplementary Figure S1 online.

Table 1 Health status scoring and description, based on antennas, eyes, rostrum, exoskeleton, uropods quality and different rate of impairments.
Full size table

Behavioral analysis

A GoPro camera was installed on top of each experimental tank to record animal behavior. Recordings of 30 min were made twice a week from 11 h to 11:30 h. An ethogram focusing on stress-related behaviors was compiled (Table 2) considering recent literature41,52. Five minutes per video per tank were analysed using Behavioral Observation Research Interactive Software (BORIS) software53. Each abnormal behavior event was tallied per tank and divided by the number of shrimps expected in the tank on that day (not counting mortalities). Prolonged surface swimming was considered effective only for a duration above 1 min. If the same individual repeated the preceding observation within 30 s, it was not counted.

Table 2 Ethogram of abnormal/stress-related behavior with description for each of them.
Full size table

Sampling

On sampling days, 12 shrimps per tank were euthanized in an ice slurry solution. Weight and length were recorded. The moulting stage was determined according to Robertson et al.54, with a stereoscope Olympus SZ model. Only shrimps in the intermolt and early premolt (C-D0-D1-D2) stages were selected for the sampling. The ventral side of the shrimp was disinfected with ethanol 70% and the hemolymph was collected with a syringe rinsed in anticoagulant (0.35 M NaCl, 0.09 M KCl, 0.03 M Trisodium citrate, 0.01 M HEPES, ph 7.5). The ratio between hemolymph and anticoagulant was adjusted to 1:1. Samples were centrifuged at 4 °C for 10 min at 800 g. The supernatant was collected, shock-frozen in liquid nitrogen, and stored at -80 °C.

The hepatopancreas was dissected and weighed. Approximately 100 mg hepatopancreas were transferred to 500 µL RNA later (Invitrogen, US), incubated for 24 h at 4 °C and stored at − 20 °C. The remaining hepatopancreas was immediately flash-frozen in liquid nitrogen and stored at − 80 °C until analysis.

Genes expression

Total RNA was isolated according to Reiser et al.55 using TRIZOL (Invitrogen), followed by RNA Mini kit extraction (Qiagen) and on-column DNase I digestion according to the manufacturer’s instructions. After extraction, total RNA content was determined with a Nanodrop ND 1000 spectrometer. Integrity was monitored for 10% of the samples using a RNA 6000 Lab Chip and the Bioanalyser (Agilent). Here, to avoid dissociation of the 28S RNA, no heating step was performed. RNA samples were diluted to 0.250 µg/µL. A premix of 4 µL RNA sample (1 µg RNA), 8.5 µL RNase-free water and 750 ng poly(dT) primer (CCTGAATTCTAGAGCTCAT)17, 500 ng/µL) was transcribed with Affinity Script Multi Temperature Reverse Transkriptase (Agilent, 200 U/µL) according to Cámara-Ruiz et al.56 in a two-step procedure. First, annealing of poly(dT) primer was established at 65 °C for 5 min, 40 °C for 3 min, 35 °C for 3 min, 30 °C for 3 min, 25 °C for 3 min and 10 °C. Then 2 µL 10*buffer, 2 µL DTT, 1 µL dNTP (10 mM each) and 1 µL Affinity Script Multi Temperature Reverse Transkriptase (RT) were added, incubated at 42 °C for 1 h and denaturation at 70 °C for 15 min. Genomic DNA contamination was monitored by a negative RT control where the RT had been replaced by DEPC treated water. Species‐specific primers (Table 3) were designed based on sequence information available in NCBI. Assay specificity was confirmed by direct sequencing (SeqLab, Germany) and monitored in each QPCR run by melting curve analysis. QPCR analysis was carried out with a CFX qPCR Cycler (BioRad). For the 25 µL PCR mix, 2 μL of the diluted samples (16.7 ng/μL) were used as a template (1.5 mM MgCl2, 200 μmol/L of each dNTPs (Qbiogene), 400 nM primer, 1*SYBR‐Green I (Invitrogen) and 1 U Invitrogen Platinum Taq polymerase). The reaction was carried out with an initial denaturation at 94 °C for 2 min, followed by 40 cycles of denaturation at 94 °C for 20 s, primer annealing at 63 °C for 20 s and elongation at 72 °C for 20 s. PCR efficiencies were determined experimentally with a dilution series of a calibrator (16.7 ng/μL). Samples were assessed in duplicate. Target gene expression was calculated by the comparative CT method (ΔΔCT) according to Pfaffl57, corrected for the assay efficiencies and normalized to actin as a housekeeping gene. Gene expression is presented as the fold increase of the expression on day 0. Even expression of actin as housekeeping gene was confirmed by comparing CT values between treatments, revealing no statistical difference (Tukey, p > 0.05).

Table 3 List of primers belonging to target genes. Lengths (in bp), code on GeneBank, efficiency (in %) and sequence (5’-3’).
Full size table

Hemolymph parameters

The hemolymph was defrosted on ice at 4 °C. Glucose and total protein in the hemolymph were determined with an automated blood analyzer (Fuji DRICHEM NX500, Tokyo, Japan), following the instructions of Weiss et al.58. Lactate was assessed following kit instructions (l-lactate, Megamix, Ireland) according to Perrot-Minnot59.

Enzyme activity

For the hepatopancreas SOD assay, a solution at 10% w:V was prepared by adding phosphate buffer (50 mM, pH 7.49). The hepatopancreas was crushed using a pestle until homogeneity was achieved. Then, it was centrifuged at 800 g for 15 min at 4 °C. The supernatant was collected and frozen at -80 °C. Activity of SOD was measured with a SOD Determination kit (no. 19160, Sigma-Aldrich, USA) using the method of Campa-Córdova et al.60. After incubation at 37 °C for 20 min, absorbance at 450 nm was measured with an Infinite 200 PRO spectrophotometer (Tecan, Mannedorf, Switzerland) on a micro-plate. SOD activity was calculated with the formula:

$$\text{SOD activity (inhibition rate \%)} = 100 \times \frac{(A_{\text{blank 1}} - A_{\text{blank 3}}) - (A_{\text{sample}} - A_{\text{blank 2}})}{A_{\text{blank 1}} - A_{\text{blank 3}}}$$

The percentage of inhibition was converted into U/mL with a standard SOD solution (no. S5395, Sigma-Aldrich, USA).

Phenoloxidase activity was measured spectrophotometrically by tracking the formation of dopachrome from L-dihydroxyphenylalanine (L-DOPA), following the method of Hernández-López et al.61, adapted for micro-plates. 10 μL plasma was added to 10 μL cacodylate buffer (10 mM, pH 7) and 10 μL L-DOPA solution (3 mg/mL in distilled water). The solution was incubated at 37 °C for 10 min. Then, 160 μL cacodylate buffer was added. The absorbance was measured at 492 nm using an Infinite 200 PRO spectrophotometer (Tecan, Mannedorf, Switzerland) at 0 and 10 min. A solution of L-DOPA and cacodylate buffer was used as negative control. PO activity was expressed in U/mL enzyme activity, defined as an increase in absorbance of 0.001 per min per mL of hemolymph.

Statistics

Statistical analysis was performed using R software (Version 4.3.2). Linearity and homogeneity of variance were checked using Bartlett and Shapiro–Wilk tests, respectively. If assumptions were not met, a non-parametric test (Kruskal–Wallis) was carried out. Where assumptions were met, data were subjected to One-Way Analyses of Variance (ANOVA). Duncan’s multiple range test was used to determine the statistical significance among groups. Data are presented as mean values and were considered significantly different if the p-value < 0.05.

Behavioral variations during both experimental phases were investigated using generalised mixed-effect models (GLMM) in the R package “glmmTMB”, using the behavior analysis of Bardera et al.62 and Bolker63 as a reference. Univariate models for each behavior type and phase were calculated using treatment and day as fixed factors and tank as a random factor. Due to the presence of zero and positive continuous responses, Tweedie distribution was used in all behavior models64. Post-hoc analyses used the R package “emmeans” for main effects. Assumptions and diagnostic value of each model were assessed using the Dharma package65.

Results

Growth performance and survival

Optimal water quality parameters were maintained for L. vannamei throughout the experiment (Table 4).

Table 4 Average and standard deviation of each water chemistry parameter during the experimental phase (n=12).
Full size table

At completion of the stress phase, significantly reduced survival rates were recorded in the high stocking density (high) group compared to the standard stocking density group (standard) and the low stocking density (low) group. The low group exhibited significantly higher mean weight, RWG, WWG, and SGR when compared to standard and high treatments. No significant differences were observed in the CF and hepatosomatic index HSI. After the recovery phase, on day 42, survival rates were comparable across treatments. However, differences in weight and total length persisted. Growth performance indices, such as SGR, RWG, WWG were significantly higher in high density compared to the low density (Table 5).

Table 5 Effects of different stocking densities (low, standard and high) on survival, growth performance, and morphological index of L. vannamei for 42 days.
Full size table

Morphological status

In stress conditions, high stocking density significantly affected the integrity of shrimp antennas and uropods (Fig. 1a and b). Shrimp in the high treatment exhibited the most severe damage to antennas (p < 0.05), whereas damage to uropods was similar between the standard and low densities. In the same experimental phase, a comparable but not statistically significant trend was observed for the exoskeleton (Fig. 1c). Rostrum and eye quality did not differ significantly between treatments (Fig. 1d and e). After the recovery phase, no significant differences were detected between treatments, and damage levels mostly decreased compared to the stress phase (Fig. 1).

Fig. 1
Fig. 1
Full size image

Impact on quality at D0, end of stress phase, and end of recovery phase (mean ± SD; n = 12) of (a) antennas, (b) uropods, (c) exoskeleton, (d) rostrum, (e) eyes.

Behavioral analysis

Aggressive behavior was observed only once and therefore, excluded from the analysis. Abnormal swimming events were significantly more frequent in higher stocking densities during the stress phase (Fig. 2a, Table 6). Similarly, high and standard groups exhibited a higher frequency of loss of balance behaviors compared to the low density (Fig. 2b, Table 6). However, the frequency of body cramps and strong escape behaviors did not vary significantly between treatments (Fig. 2c, Table 6). During the stress phase, a day effect was detected for loss of balance and body cramping behavior. (Fig. 2, Table 6). Abnormal behavior occurrences declined during the recovery phase, with a day effect observed for loss of balance behavior (Fig. 3, Table 6). No statistically significant differences were detected among treatments for any behavior type during recovery (Fig. 3, Table 6).

Fig. 2
Fig. 2
Full size image

Predicted lines with 95% CI and data points of observed values (dots) representations (n = 4). The number of occurrences of abnormal behavior events per shrimp present in the tank on that day over 6 different time points during the stress phase. (a) Abnormal swimming events, (b) loss of balance, (c) body cramps, (d) strong escape behaviors.

Table 6 Results of GLMM used to determine the effects of stocking density treatments, day of observation on L. vannamei behaviors.
Full size table
Fig. 3
Fig. 3
Full size image

Predicted lines with 95% CI and data points of observed values (dots) representations (n = 4). The number of occurrences of abnormal behavior events per shrimp present in the tank on that day over 6 different time points during the recovery phase. (a) Abnormal swimming events, (b) loss of balance, (c) body cramps, (d) strong escape behaviors.

Hemolymph parameters

Hemolymph analysis revealed no significant variation in glucose, l-lactate, or total protein concentrations across treatments during both phases (Fig. 4). There was no temporal difference within treatments detected.

Fig. 4
Fig. 4
Full size image

Concentration or activity values (n = 12) in shrimp hemolymph at D0, end of stress phase, and end of recovery phase of (a) glucose, (b) total proteins, (c) l-lactate, (d) phenoloxidase activity.

Enzyme activity

Phenoloxidase activity in hemolymph and superoxide dismutase activity in hepatopancreas tissue showed no significant differences among treatments after both stress and recovery phases (Fig. 5).

Fig. 5
Fig. 5
Full size image

Superoxidase activity (n = 12) in shrimp hepatopancreas at D0, end of stress phase (day 21), end of recovery phase (day 42).

Gene expression

Gene expression analysis revealed significant effects of high density on Gpx and HSP70 after the stress phase, with both markers showing higher expression levels in the high density group compared to standard and low densities (Fig. 6a and c). Similarly, HSP90 in high and standard stocking density exhibited higher expression than low, during the stress phase. Lysozyme expression did not differ statistically between treatments (Fig. 6f), no distinct patterns were observed for Tgf-ß, akirin and p53 during the stress phase (Fig. 6b, e and g). After the recovery phase, gene expression levels returned to baseline, with no differences among treatments (Fig. 6).

Fig. 6
Fig. 6
Full size image

Normalised gene expression (mean ± SD; n = 12) of (a) glutathione peroxidase, (b) p53, (c) heat shock protein 70, (d) heat shock protein 90, (e) akirin, (f) lysozyme, (g) transcriptions factor ß, at D0, end of stress phase and end of recovery phase.

Discussion

This study applies a multidisciplinary approach on the stress response cascade of L. vannamei under chronic crowding stress conditions. Beyond evaluating survival, growth metrics, biomarkers for cellular stress, modulations of the immune system and morphological alterations, it provides valuable first-time insights into injury frequency and stress-related behaviors. A morphological index and an ethogram have been designed through literature review and applied, while quantitative and qualitative manual observations have been conducted using open-source software. Additionally, the recovery phase sheds light on the capacity of this species to cope with density stress, heal from injuries, and evaluate eventual compensatory growth and stress plasticity.

As previously observed by da Silveira et al.66 and Liu et al.67, survival rates in shrimp are strongly affected by stocking density. In the present experiment, elevated stocking densities increased mortalities, clearly indicating suboptimal conditions on a systemic scale. Nevertheless, mortality normalised almost immediately when the recovery phase began, resulting in high survival across all treatments. Similarly, the growth indices analysed in this study indicated faster growth under low-density conditions, consistent with previous findings67,68. Among the treatments, during the stress phase, low density was the only condition that resulted in increased biomass, clearly suggesting that the high and standard groups were affected by suboptimal conditions. Growth and survival reduction are generally considered to be effective indicators of chronic stress. Increasing densities can induce competition for resources such as feed, space occupation (substrate mainly as benthic species) and vulnerability during the moulting phase62,69,70,71. In our experimental system, the great majority of exuviae and mortalities were recorded and collected in the morning. At that time and during usual feeding and tank maintenance, no aggressive behavior was observed. Increased cannibalism may have occurred, in particular at high densities or at crepuscular/night periods. For commercial farming, maintaining low stocking densities is not economically viable. Feeding protocol optimisation as increased feeding frequency and a well spread distribution could reduce dominance hierarchies and cannibalistic behavior62,72. Production strategies must strike a balance to optimize yield within the shortest production cycles. In crowded systems, increased pathogenic bacterial load may impair growth and health in shrimps as shown in ponds and biofloc systems73,74. Proper disinfection, constant water/animal monitoring, use of probiotics might be helpful tools to limit pathogenic and opportunistic bacterial infections. Tailored solutions are essential, taking into account farm designs and husbandry practices. At a commercial level, mitigating these issues may involve adjustments in feeding frequency and distribution, photoperiod control, the provision of additional shelters or enrichments, and optimized aeration strategies.

Compensatory growth is a well-documented phenomenon in L. vannamei, observed after stress, such as nitrate exposure and high stocking densities50,75,76. At the end of the experiment, the growth rates of the shrimp kept in high density significantly exceeded the standard and low groups, reaching values in line with growth performance of the low stocking density during the stress phase. This highlights the capacity of L. vannamei to retrieve optimal growth after crowding chronic stress. Additionally, there was partial compensatory growth under recovery densities, as they did not completely close the weight gap of the former low stocking density treatment. The observed recovery in growth rates aligns with the findings from Barajas-Sandoval et al.35, where compensatory growth was observed after low temperature exposition and short fasting period.

Chronic stocking density stress significantly affected shrimp health status, most prominently the damage to antennae and uropods. These injuries are well-documented in the literature, particularly under intensive farming conditions, and are supported by several studies43,77. The primary causes are attributed to frequent abrasion of appendages against hard physical substrates, interactions with other shrimps (stepping on each other), and increased antagonistic behaviors. Visible carapace damage particularly concerns shrimp farmers in Europe, where whole, unpeeled shrimps from RAS-based production are marketed at a premium price. Long, undamaged antennae and a pristine exoskeleton are desirable for such a market segment. In particular, melanosis could render the shrimp unsaleable and result in economic losses for producers. No observable variations were noted in the condition of the eyes or rostrum under these stocking density conditions, suggesting that these body parts remain unaffected. After the recovery period, shrimp exhibited notable biological improvement with healing of damaged appendages. Minor injuries, such as broken appendages, typically heal after one or two moulting cycles, although the precise timeframe remains uncertain and likely varies with the extent of the damage. Future research should focus on the regenerative capacity of shrimp and its relationship with moulting events, as well as welfare assessment using indices inspired by the health status guide proposed here, both ideally through weekly assessments18,78,79,80.

Aggressive behaviors were not detected during experimental observations, though it is hypothesized that such interactions may have occurred nocturnally, coinciding with moulting events and mortalities recorded here. Indeed, during moulting, shrimp are more vulnerable to physical damage, increasing the likelihood of cannibalism70. Cramping and strong escape behaviors are not mentioned to be correlated with crowding41,81,82. Movement patterns, such as swimming abnormalities and loss of balance, exhibited a direct correlation. Specifically, swimming abnormalities consisted of irregular movement patterns and prolonged slow surface swimming, both of which are recognized in the literature as indicators of poor shrimp health41,79. Loss of balance was mainly observed for shrimps lying on a side on the tank bottom and could be associated with moribundity, as indicated for thermal stress or for a prolonged and problematic moulting event83,84,85. During recovery, behavioral abnormalities also declined remarkably, with fewer events recorded and a consequent decline in the frequency of occurrence. These findings provide a foundation for further research on the behavioral effects of chronic stress. Targeted studies would enhance our understanding of these phenomena and their implications for shrimp welfare and recently developed artificial intelligence-driven production management86.

The minor response in commonly applied hemolymph parameters of the secondary stress response is surprising considering the previously described consequences of increasing density on survival, growth performance and health status. Indeed, suboptimal conditions due to ammonia stress or handling led to hyperglycemia with increased lactate and protein levels87. However, in the long-term, persistent distress adaptive responses, such as glucose or lactate increases, will most likely return to near baseline concentrations60,87,88,89,90. To explain this phenomenon, homeostatic mechanisms facilitate the mobilization of glycogen, carbohydrates, and lipids from storage organs, such as the hepatopancreas, to release readily available energy molecules15,89. Further analysis of storage organs would be valuable in confirming homeostasis in response to long-term stresses.

In the current study, prevailing alterations of the immune system were minor. Indeed, stable phenoloxidase activity and Akirin expression were observed under persistent stress in all experimental groups. PO activity is reported to decrease under crowding stress after both short-term (12 h) and long-term (8 weeks) exposure47,91. The discrepancies correlated with those studies might be explained by the use of a different culture system91, homeostatic adjustments or variation in shrimp size, as larger specimens are reported to be more resistant to stress92,93. Akirin activation doesn’t seem to be associated with crowding, but only with increased expression during bacterial infection29. Non-significant increases in lysozyme were observed with increasing density in the current study, aligning with findings of Yuan et al.94 and Liu et al.95. TGF-β signalling did not increase during the stress phase, again suggesting a mild influence of crowding on the immune system. The TGF-ß cytokine exhibits anti-inflammatory effects and regulates the apoptosis pathway. TGF-ß was increased in previous studies on salinity and temperature stress96,97. An expanded analysis of immune parameters, such as hemocyte count or hemolymph clotting time88 could provide deeper insights into this topic. The absence of major alterations in the immune response is complex to interpret in the context of the sophisticated capacity of invertebrates to react to stressful environments. It may be attributed to either a lack of sensitivity to the stressor, homeostatic adaptation or inhibited expression due to distress94. In this context, genetic improvements and farming adaptations of the species may have resulted in more resilient individuals, that cope better with stress, as suggested by Shi et al.98 and Wang et al.99. An expanded analysis of immune parameters, such as hemocyte count or clotting time of hemolymph88 could provide deeper insights in the future.

SOD activity in the hepatopancreas appears unaffected by crowding stress, with no significant changes in activity. This aligns with studies on biofloc systems and density related experiments60, however other studies evidence variation, maybe due to shrimp size or assay methodology reasons67,100. Analysing SOD activity in other tissues, such as the hemolymph and muscle, or different shrimp sizes may reveal more treatment-specific variations, as observed by Lin47 and Campa-Cordova60. Contrarily, Gpx appears to respond to intense crowding events with increased expression, similar to its upregulation under bacterial challenge and pH stress92,101. During recovery, Gpx expression returns to basic levels similar to the other parameters observed. Unlike under acute stress exposure such as hypoxia and ammonia/nitrite102,103, p53 expression did not variate in both phases. Based on these results, p53 might not have a key role in chronic stresses, being not responsive over prolonged stress or with other factors involved in regulating cell processes as apoptosis. Alternatively, the mild oxidative stress and consequent cellular damage might have not been strong enough to enable its pathway. During the stress phase, HSP70 and HSP90 increased with rising crowding. Similar to other studies104,105, HSP70 seems to be a biomarker of density-related stress. Although HSP90 has been less investigated than HSP70, it is reported to be altered at lower temperature, salinity and ammonia106,107. Based on the current results and its sensitivity to crowding, HSP90 seems to be a valid general environmental stress marker. As for Gpx, HSP70 and HSP90 expression returned to basic levels after recovery. Overall, in contrast to the immune response, oxidative stress markers appear to be highly responsive to the examined stressors, even when the stress level seems mild.

Conclusions

Increased stocking density adversely affected shrimp survival, growth, and morphological status, while behavioral analyses revealed increased frequencies of abnormal movement patterns. Homeostatic mechanisms probably from storage organs helped maintain constant key hemolymph parameters. A mild upregulation of commonly assessed markers indicating cellular/oxidative stress was evidenced. Effects on the immune system indicated a lack of sensitivity for crowding or distress levels. During the recovery phase, shrimps recovered fast from injuries and stress and showed partial compensatory growth, offering insights into management of animals after stress exposure. Health status and behavior data could serve as a base for developing welfare indices or AI-based monitoring systems at a commercial level.