Introduction

Aquaculture is essential to global food production (Garlock et al. 2020; Naylor et al. 2021; Tacon 2020). The Greenshell™ mussel, Perna canaliculus, is the most valuable aquaculture species in New Zealand with approximately 97,000 t harvested each year, creating a revenue of NZD$ 381 million annum−1 (USD$ 235 million annum−1) (Stenton-Dozey et al. 2021). One of the major issues facing mussel farmers in New Zealand is high loss of spat during the nursery stages of the aquaculture production cycle in the first few months following seeding (Skelton and Jeffs 2020; South et al. 2019). The spat used in the New Zealand aquaculture system are predominantly obtained from the wild, and their abundance is limited by quota regulations and the highly variable timing of spatfall (Jeffs et al. 1999; South et al. 2020). Losses of spat during the nursery stages are a barrier to increasing production and can also have direct impacts on the productivity of the industry (Skelton et al. 2022). The causes of spat loss are, however, not well understood and could be driven by a suite of environmental (e.g. water temperature, food availability) and procedural factors (e.g. transfer duration, seeding density) that can elicit a range of physiological and behavioural stress responses (Delorme et al. 2021a, b; South et al. 2021). Accordingly, understanding the stress responses of spat in relation to procedural aspects of the nursery cycle is a high priority for mussel farmers who aim to manage and optimise mussel production.

A commercial Greenshell™ mussel farm is typically 3 ha and consists of 10 longlines made up of a double backbone system supported by buoys, with culture ropes seeded with mussels continuously looped over the backbones and extending 5–10 m deep (Jeffs et al. 1999; Stenton-Dozey et al. 2021). Beach-cast spat (250 μm–10 mm) use seaweeds and hydroids as a settlement substrate and are sourced from a single beach in the far north of the country. The beach-cast material containing the spat are then seeded by deploying the substrates alongside polypropylene culture ropes and using a mesh stocking to hold them in place (Jeffs et al. 1999). Over time, spat migrate to the culture rope, stocking, or elsewhere (Skelton and Jeffs 2020; South et al. 2017). Losses of cultured spat are common after their initial seeding (the nursery stage), when up to 99% of spat can be lost (Skelton and Jeffs 2021; South et al. 2020). Mortality and secondary settlement can contribute to spat loss; however, their drivers are complex and difficult to predict (Carton et al. 2007; Hayden and Woods 2011; South et al. 2020). Once mussels have reached 10–20 mm in shell length, they are stripped from the culture rope and re-seeded at a lower density, a process called inter-seeding (Jeffs et al. 1999; South et al. 2020). Because spat loss is greatest in the nursery stage of cultivation, optimization of the proportion of spat that remain at inter-seeding will enhance the harvestable yield of adult mussels.

Marine mussels among other shellfish are exposed to natural fluctuations of environmental conditions such as changes in temperature, salinity, and food availability (Bayne and Bayne 1976; Shumway and Parsons 2016). Additionally, mussels can also be affected directly by anthropogenic activities such as handling, transportation, and other farming procedures that stem from cultivation such as overcrowding due to high densities (Calderwood et al. 2014; Cubillo et al. 2015). Due to the limited ability to control biotic and abiotic factors in sea-based aquaculture systems coupled with the high demand for an abundant and quality product, it is critical to understand the effect of farming-induced stressors such as seeding density, which can result in poor animal performance or even death. Modifying seeding density is considered a potential method to manage spat losses in Greenshell™ mussel cultivation as it can alter biofouling accumulation and mitigate overcrowding on the culture ropes (Bordignon et al. 2021; Cubillo et al. 2015); however, high seeding density in mussels (Mytilus galloprovincialis and Mytilus edulis) aquaculture can increase competition for resources and has been shown to impact survival and growth negatively (Cubillo et al. 2012; Guiñez 2005; Karayücel et al. 2015; Lauzon-Guay et al. 2005a, b).

Molecular, biochemical, and physiological biomarkers can vary when marine invertebrates are exposed to stressful conditions such as changes in temperature, pH, food availability, and overcrowding due to high stocking densities (Bayne and Bayne 1976; Delorme et al. 2021a; Gao et al. 2017; Gazeau et al. 2013; González Durán et al. 2018; Gosling 2021; Hawes et al. 2018; Leung et al. 2020; Liu et al. 2019; Marčeta et al. 2020; Marigomez et al. 2017). Stress-on-stress interactions can occur when one stressor reduces the organism’s ability to cope with an additional stressor (Delorme et al. 2020). This study focused on evaluating how variation in seeding density affected stress biomarkers (lipid peroxidation [LP] and total antioxidant capacity [TAC]) in commercially grown Greenshell™ mussel spat, using a spatio-temporal experimental design. Spat viability was also measured at the time of sampling using freshwater immersion tests to determine the physiological condition of mussels at the time of sampling from mussel farms. Understanding how seeding density impacts stress biomarkers in Greenshell™ mussels may help to implement seeding techniques to minimise mussel stress and mitigate stress-on-stress interactions.

Materials and methods

Experimental design

Greenshell™ mussel spat were obtained from three 10 kg bags of beach-cast material harvested from Ninety Mile Beach, north-western New Zealand in September 2019. Beach-cast material was transported in a refrigerated truck (6–7 ℃) to central New Zealand (Havelock), where the three bags were amalgamated to produce 30 kg of spat material for the experiment. Five samples of the spat material (37–54 g) were haphazardly removed prior to seeding and frozen at − 20 °C. Samples were weighed, washed, and counted under a dissecting microscope. Mean initial mussel density was 156 ± 15 mussels g−1 of beach-cast material (± standard error, SE). Seeding of spat was done at five conventional longline mussel farms at each of two sites in the outer Marlborough Sounds, central New Zealand. Seeding of the mussels onto the culture ropes used a purpose-designed vessel following local farming procedures. Mussels were seeded onto the culture ropes at three density treatments: low (257 mussels 10 cm−1), mid (515 mussels 10 cm−1), and high (1030 mussels 10 cm−1). All farms at both sites held a 16 m loop of culture rope for each seeding density, with ropes spaced 0.5 m apart (Fig. 1). This study occurred during the nursery stage of the aquaculture production for the Greenshell™ mussel.

Fig. 1
figure 1

Experimental set-up on a long-line mussel farm replicated at five farms at each of two sites in the outer Marlborough Sounds. At each mussel farm, there were three, 16 m loops of culture rope of low-, mid-, and high-seeding density. The culture ropes were spaced 0.5 m apart, and 10 cm sections of rope were collected from 0.5 m, 2.5 m, 4.5 m, and 6. 5 m from both the left and right side of each loop of culture rope in December 2019 and February 2020

Sample collection

Samples were collected in December 2019 and February 2020 to determine the total antioxidant capacity (TAC), lipid peroxidation (LP), and viability of Greenshell™ mussels at the three seeding densities (low, mid, and high). Additionally, at each seeding density, samples were taken from each site, farm, and from both the left and right sides of each loop of culture rope at four depth strata: 0–1 m, 2–3 m, 4–5 m, and 6–7 m (hereafter 0.5 m, 2.5 m, 4.5 m, and 6.5 m, respectively; Fig. 1). Samples for viability assays were taken at all seeding density, depth, farm, and side of loop combinations at each site, whereas samples for TAC and LP assays were taken from only three farms at each site at 0.5 m and 6.5 m depth from only one side of each loop of culture rope. Some sections of the culture ropes at site 2 were lost due to rough weather between the December 2019 and February 2020 sampling dates resulting in loss of samples at 2.5 m and 6.5 m depth at one farm. For TAC and LP analyses, supplementary samples were taken from a fourth farm at site 2 in February 2020.

At each sampling event, experimental culture ropes were lifted from the water, and a 10 cm section was cut at each depth. Then the remaining culture rope was reconnected and returned to the water once all samples from each culture rope were collected. From the 10 cm rope section, 2–6 mussels were collected from designated samples for TAC and LP analyses. Mussels were dissected in situ, and the excised tissue was placed into 3.6-mL cryogenic tubes and snap-frozen in liquid nitrogen. Samples were transported to the laboratory and stored at − 70 ℃ until analysis. The remaining 10 cm rope sections were labelled and placed in a zip-lock bag and in a cold chiller bin for viability tests. A total of 36 samples for both TAC and LP were collected at each sampling time (December 2019 and February 2020), and a total of 240 and 192 samples were collected for viability tests in December 2019 and February 2020, respectively.

Oxidative stress biomarker assays

For TAC and LP analyses, the frozen mussel tissues collected were individually ground and homogenised into a frozen powder using a pre-chilled (LN2) mortar and pestle and weighed (to the nearest 0.1 mg). For the TAC assay, an aliquot of 100 mg ± 10 mg was collected from the homogenised tissue powder for each sample, while for the LP assay, an aliquot of 20 mg ± 3 mg was collected for each sample. Aliquots of frozen powder were placed in 1.5 mL Eppendorf tubes.

Total antioxidant capacity assay

Total antioxidant capacity (TAC) was measured using an Antioxidant Assay Kit (CS0790, Sigma-Aldrich®) following manufacturer instructions. For the assay, the powdered samples were homogenised using assay buffer solution and glass beads (Qiagen 5 mm) in a Qiagen TissueLyser II for 5 min. The homogenate was then removed from the tube, placed into 2-mL Eppendorf tubes and centrifuged at 12,000 × g for 15 min at 4 ℃. The supernatant from each tube was removed and frozen in aliquots at − 70 ℃. A clear 96-well plate was used for the assay, with 10 μL of supernatant added to each well in duplicate, together with the reagents specified in the protocol. Where required, samples were diluted fivefold in assay buffer solution to standardise the range of possible absorbance values. The TAC values were then obtained from reading the absorbance at 405 nm using a BioTek Synergy 2 plate reader and Gen5 3.05 software. The TAC concentration (mM) was calculated from Trolox ([ ±]-6-hydroxy-2,5,7,8-tetramethylchromane-2-carboxylix acid) standard curves (0–0.42 mM, mean (± SE) R2 = 0.712 ± 0.059) and expressed as Trolox equivalent concentration. A greater Trolox concentration indicates greater TAC, which demonstrates that the mussel is better prepared to resist oxidative stress.

Lipid peroxidation assay

Lipid peroxidation (LP) was measured using the Lipid Peroxidation (MDA) Assay Kit (MAK085, Sigma-Aldrich®), following manufactures’ instructions and adjustment of the volume of reagents according to the tissue mass in each sample. Briefly, each sample was homogenised in 600 μL MDA Lysis buffer containing 6 μL of butylated hydroxytoluene (100 ×). The samples were then centrifuged at 13,000 × g for 10 min at room temperature. Then, 200 μL of the supernatant from each sample was transferred to a microcentrifuge tube with 600 μL of TBA and incubated at 95 ℃ for 60 min. Once cooled to room temperature, 200 μL of the supernatant and TBA mixture was added into 96-well plates in triplicates, and the absorbance was read at 532 nm using a BioTek Synergy 2 plate reader and Gen5 3.05 software. The MDA concentration (nmol mL−1) was calculated from MDA standard curves (0–20 nmol, mean [± SE] R2 = 0.991 ± 0.028). A greater MDA concentration indicates greater lipid peroxidation which suggests that the mussel is under greater oxidative stress.

Viability test

To determine viability, mussels were exposed to a freshwater immersion on the day following sample collection. During freshwater immersion, mussels that are in a good physiological condition will respond to the osmotic shock by closing their valves. Mussels in a poor physiological condition or dead do not respond (Webb and Heasman 2006). Thus, during a freshwater immersion test, mussels can appear either ‘viable’ (closed valves and alive and good physiological condition at the time of collection) or ‘non-viable’ (open valves or empty and dead at the time of collection). To perform the freshwater test, all contents from the 10 cm culture rope samples were removed and rinsed through a 1000 μm sieve. All contents retained in the sieve were immersed in freshwater, and the number of open and empty mussels was counted after 30 s immersion. Samples were then placed at − 20 °C and stored until the total abundance of mussels in each sample was quantified, and the proportion of closed, open, or empty mussels at the time of collection was calculated.

Statistical analysis

To examine the impacts of the factors manipulated or measured, linear and generalised linear mixed models were fit using R package lme4 (Bates et al. 2011) in RStudio version 1.4.1106 (RStudio Team 2021). Analyses included four fixed effects: seeding density with three levels (low, mid, and high); depth with two levels for TAC and LP (0.5 m and 6.5 m) and for viability depth was considered as a continuous factor with four ordinal values: 0.5 m, 2.5 m, 4.5 m, and 6.5 m; site with two levels (site 1 and site 2); and time with two levels (December 2019 and February 2020), as well as the interactions between all combinations of fixed effects. There were three random effects in TAC and LP statistical analyses: farm, rope, and sample ID, to account for the variation among farms, culture ropes, and sample replicates (i.e. the duplicates or triplicates of each sample), while viability analysis had farm and rope as random effects. Due to the number of interactions among the fixed-effects model, convergence and singularity warnings were common using the default optimizers. Accordingly, all models were fit with all available optimisers using the allFit function of lme4 and the parameter estimates examined for consistency among different optimizers. Overall, parameter estimates were consistent across all available optimizers and considered robust.

For the TAC assay, the response variable was the mean total antioxidant concentration (mM Trolox equivalent), and for the LP assay, the response variable was the mean MDA concentration (nmol mL−1), and a Gaussian error distribution was assumed. For viability, a binomial error distribution and a logit link were assumed, where the response variable the number of viable (closed valves) and non-viable (open valves or empty shells) mussels in each sample. For mussel viability, the time between exiting the ocean and the freshwater immersion test was recorded as ‘aerial exposure’ (10–36 h) for each sample and added as an offset in viability analysis given the strong relationship between aerial exposure and viability (F1 = 82.72, p < 0.001). To determine statistical significance of our fixed effects, backward model selection by single-term deletions was performed, using the drop1 function in the R package lmerTest (Kuznetsova et al. 2015) (Appendix Table  1). Tests were only produced if a fixed effect was not involved in a significant (i.e. p < 0.05) higher-order interaction.

Results

Seeding density had inconsistent effects on total antioxidant capacity (TAC) across depths, sites, and times (Fig. 2A–D; Appendix Table 2: Site×Density×Depth×Time). In December 2019, at the low seeding density, TAC decreased with increasing depth by 16% at site 1 (Fig. 2A) and 52% at site 2 (Fig. 2B). In December 2019, TAC increased by 28% with increasing depth at the high seeding density at site 2 (Fig. 2B). In February 2020, TAC was impacted by seeding density at site 2 only, where TAC increased by 77% with increasing depth at the low seeding density (Fig. 2D); however, on average, TAC was 71% greater in December 2019 compared with February 2020 (Fig. 2A–D). Lipid peroxidation (LP) was only significantly impacted by time (F1,210 = 5.656, P = 0.018), where LP was, on average, 25% greater in February 2020 compared with December 2019 (Fig. 3A–D).

Fig. 2
figure 2

Total antioxidant capacity (TAC, mM Trolox equivalent) in Greenshell™ mussels (P. canaliculus) at three seeding densities (low: 257 mussels 10 cm−1 of culture rope; mid: 515 mussels 10 cm−1 of culture rope; high: 1030 mussels 10 cm1 of culture rope), two depths (0.5 m and 6.5 m), at site 1 and site 2 in the Marlborough Sounds in December 2019 and February 2020. Data are means ± standard error (SE)

Fig. 3
figure 3

Total lipid peroxidation (MDA concentration [nmol mL−1]) in Greenshell™ mussels (P. canaliculus) at three seeding densities (low: 257 mussels 10 cm−1 of culture rope; mid: 515 mussels 10 cm−1 of culture rope; high: 1030 mussels 10 cm1 of culture rope), two depths (0.5 m and 6.5 m), at site 1 and site 2 in the Marlborough Sounds in December 2019 and February 2020. Data are means ± standard error (SE)

Accounting for the effect of aerial exposure, viability varied among sites, seeding densities, and depths (Fig. 4A–D; Appendix Table 2: Site×Density×Depth). At mid- and high-seeding densities at site 1, viability was up to 3% greater at 6.5 m compared with 0.5 m (Fig. 4A, C). At site 1, viability was slightly greater (approximately 1%) at mid- and high-seeding density compared with low-seeding density (Fig. 4A, C); however, at site 2, seeding density and depth had a lesser impact on viability (Fig. 4B, D). In December 2019, viability had a positive relationship with seeding density; however, in February 2020, there was a slight negative relationship between seeding density and viability (Fig. 4A–D; Appendix Table 2: Density×Time). Additionally, the effects of depth on viability were more apparent in December 2019, where viability was typically greater in samples collected at deeper depths compared with shallower depths at site 1, but lower at deeper depths at site 2 (Fig. 4A, B; Appendix Table 2: Site×Depth×Time). Differences in the effects of depth on viability between sites were reduced in February 2020 (Fig. 4C, D). Despite small differences in viability among the experimental treatments, viability of mussels remained high across all treatments (> 90%; Fig. 4).

Fig. 4
figure 4

Viability of Greenshell™ mussels (P. canaliculus) spat at three seeding densities (low: 257 mussels 10 cm−1 of culture rope; mid: 515 mussels 10 cm−1 of culture rope; high: 1030 mussels 10 cm1 of culture rope), four depths (0.5 m, 2.5 m, 4.5 m, and 6.5 m), at site 1 and site 2 in the Marlborough Sounds in December 2019 and February 2020. Data are means ± standard error (SE)

Discussion

While mobile organisms can relocate to more favourable conditions, sessile or sedentary species have limited ability to escape stressful conditions and must either tolerate environmental challenges or perish. Tolerating challenging conditions requires either a physiological or whole-organism response. Stress biomarkers in mussels can be investigated to understand physiological response to stressful conditions (Collins et al. 2020; Delorme et al. 2021a; Giannetto et al. 2017). This study showed that seeding density had no strong effects on total antioxidant capacity (TAC) or lipid peroxidation (LP). There was, however, an effect of time (early vs late summer) on levels of these biomarkers in Greenshell™ mussel spat.

Greater oxidative stress levels in spat at the end of the experimental period coincided with the peak of the austral summer. Water temperature in central New Zealand (Marlborough Sounds) is highest in February and March, where it can be ~ 1 ℃ warmer than in December (Broekhuizen et al. 2021; Hickman et al. 1991). Summer warming during this study followed the long-term pattern in this region (New Zealand King Salmon 2020), indicating that summer water temperatures in 2019–2020 were representative of past growing conditions when summer mortality has not been observed. In addition to increases in temperature, increases in spat size and thereby competition for space and food, as well as decreases in Chlorophyll a, particulate carbon, and organic matter in central New Zealand over summer may have affected levels of oxidative stress in our study (Hickman et al. 1991). Therefore, although a significant increase in stress during the peak of summer was detected, given the overall high viability of mussels sampled, the increased levels of LP observed in February could be considered within the normal range for spat in the field and may serve as a baseline for future studies when marine heat wave events do occur.

Seeding density is a variable that can be manipulated by farmers and can have a variety of benefits. Nevertheless, because almost all spat used in commercial mussel aquaculture in New Zealand is sourced from wild beach-cast material, where spat density can vary by 1000% between harvests (Jeffs et al. 1999; Skelton and Jeffs 2022), farmers use their own experience and judgement when seeding culture lines. Seeding densities in this study were low but reflected the operational seeding density for the batch of spat available for this study (i.e. 10 kg beach-cast material per 150 m of rope). In Skelton and Jeffs (2022), for example, spat collected one year prior to our study were seeded at up to 59,000 m−1 of culture rope (Skelton and Jeffs 2022), five times higher than our high-density treatment. While Skelton and Jeffs (2022) found losses of mussels to be much greater at high density which could suggest survival was lower, survival in response to density was not directly measured. Additionally, Lauzon-Guay et al. (2005b) found that survival of blue mussels (Mytilus edulis) only significantly decreased with increasing seeding density when the spat were of a smaller size (13.34–14.69 mm) compared to larger size spat (> 20.81 mm). In these cases, survival was up to 43% greater in the low seeding density (91 mussels 10 cm−1) compared to the high seeding density (162 mussels 10 cm−1) (Lauzon-Guay et al. 2005b). Similarly, the greatest difference in viability between seeding densities was at site 1, in December 2019 at 6.5 m depth where viability at the low seeding density was 7.8% greater than at the high seeding density. However, the relatively high viability (> 90%) of spat across all treatments suggest that the oxidative stress levels were non-detrimental to the spat performance in this study and that spat were in relatively good physiological condition even at the high seeding densities. It would be interesting to examine the oxidative response of mussels and viability to a wider range of seeding densities and seeding densities higher than those considered here to develop best-practice seeding density guidelines.

While both TAC and LP indicate levels of oxidative stress, LP captures the effects of reactive oxygen species (ROS) on polyunsaturated fatty acids critical for growth in juvenile bivalves, but is also related to how metabolically active an organism is, which may not necessarily mean the organism is under stress (Gallager et al. 1986; Lulijwa et al. 2021; Lushchak 2011). Differences in the conditions experienced by mussel spat among sites and depths influenced TAC, but LP in this study indicated that although factors such as temperature, oxygen, and nutrient availability (Dowd et al. 2013; Nogueira et al. 2017; Viarengo et al. 1995; Wilhelm Filho et al. 2001) may have varied among sites and depths causing oxidative damage, the effects of ROS on lipid production was unaffected by these experimental treatments. Delorme et al. (2020) showed that increased fasting duration resulted in increased oxidative damage and a reduced ability to enhance antioxidant activity in spat. Additionally, tolerance to additional stress decreased with greater fasting, indicating that prolonged periods of low food availability can have detrimental effects on mussel performance (Delorme et al. 2020; Dowd et al. 2013). The effects of time on both TAC and LP suggest that as mussels grew, greater intraspecific competition may have resulted in lower food availability enhancing oxidative damage to lipid production pathways. Importantly, however, changes in antioxidant capacity during our experiment was only impacted by seeding density differently at the various depths, sites, and times, indicating that although stress levels may have been higher at the end of our experiment, greater seeding density did not significantly influence the stress levels in mussel spat.

The reliable supply of wild Greenshell™ mussel spat remains a major bottleneck for the industry in New Zealand. If the industry continues to grow while relying on wild caught spat, farm methods must adapt to make the most efficient use of the spat resource. Given that the nursery stage is the period of the production cycle with the greatest losses, focusing efforts on minimising spat losses will result in the greatest return for research effort. Furthermore, rather than relying on whole organism responses as indicators of production performance, physiological assays could provide diagnostic tests of stress (Delorme et al. 2021a) and potentially give farmers the opportunity to develop intervention measures to minimise or prevent large-scale losses and enhance overall production of Greenshell™ mussel spat.