Evaluation of three common methods of bulk lipid quantification in soft tissues of marine benthic invertebrates

Comparative analyses of bulk lipid content were carried out using three macrobenthic invertebrate species, representing various taxonomical classes and of different lipid compositions and contents in their soft tissues: the common ragworm Hediste diversicolor (O.F. Müller 1776) (Polychaeta), the mussel Mytilus trossulus (Gould 1850) (Bivalvia) and the brown shrimp Crangon crangon (Linnaeus 1758) (Malacostraca). The animals were sampled in the shallow (water depth down to 9 m) coastal zone of the Gulf of Gdańsk (southern Baltic Sea; Fig. 1). The ragworms were collected using a van Veen grab (catch area 0.1 m²) at one site in the mouth of the Wisła Śmiała River (VIST) (φ 54°21’50.90“N, λ 18°47’00.50“E) at depth of 9 m in May 2017. The mussels were collected with a standard benthic dredge at one coastal site Mechelinki (MECH) (φ 54°36’26,40“ N λ 18°31’35.40“ E) in May 2018, and the shrimps were sampled using a landing net from the seashore at sandy beach adjacent to Gdynia harbour (GDY) (φ 54°29’54.73” N λ 18°33’43.73” E) in April 2018.

Figure 1

Location of sampling sites in the Gulf of Gdańsk.

Special care was taken to select individuals of a limited length range i.e. H. diversicolor (42–46 mm), M. trossulus (28–41 mm) and C. crangon (26–39 mm) as size has been demonstrated to sometimes affect biochemical constituents and physiological condition of marine benthic invertebrates (Sukhotin et al., 2003). Immediately after sampling, live individuals were transported to the laboratory in containers filled with seawater collected in situ and frozen in separate vials at -80°C. Before freezing, the soft tissues of each bivalve were removed. Frozen organisms and soft tissue were lyophilised individually in the STERIS LYOVAC GT2 (Germany) freeze dryer at room temperature and 0.03 mbar pressure for 48 hr and then homogenised manually in a porcelain mortar. The homogenises of all individuals belonging to one species were then combined to create three aggregate samples (pools), each containing 30 individuals of polychaetes, 26 individuals of bivalves and 33 individuals of crustaceans.

Before the analysis, thick-walled SCHOTT DURAN (Germany) tubes were quenched several times in boiling and cold water (0°C). All test tubes and glass balls were washed with dishwashing liquid, rinsed with distilled water and dried at room temperature. To ensure that all residual lipids were removed, the tubes were rinsed with chloroform and dried in a laboratory air dryer at 45°C (Christian, 2009).

The same extraction method (Bligh & Dyer, 1959) was used for all lipid determination methods. A weight of 10 ± 0.1 mg of homogenised tissue (in 20 replicates for each species and each method, n = 20), 1 cm³ of pure chloroform and 2 cm³ of methanol (≥99.6%) were added to a thin-walled glass test tube, and the mixture was centrifuged for 10 min (3000 rpm, 10°C). The supernatant was then transferred into new thin-walled glass test tubes or thick-walled glass test tubes in the case of the Marsh and Weinstein method. The lipids remaining in the pellet were re-extracted using the same procedure, and the supernatant was collected again in the same test tube. Next, 4 cm³ of distilled water was added to the supernatant derived from the double extraction. The two-phase mixture was separated by centrifugation for 10 min (3000 rpm, 10°C). The upper layer, which contained a mixture of methanol and distilled water, was removed carefully with an automatic pipette. Evaporation of the lower layer, which contained lipids dissolved in chloroform, was carried out in a laboratory air dryer at 45°C for 24 hr. After drying, white residue was visible at the bottom of the test tubes.

Routine quality control validation parameters and the time required for completing analysis (used here as a proxy of sample throughput) were determined for each method as factors that allowed evaluation of the lipid determination methods. In order to assess the efficiency of each method, precision (CD), limit of detection (LOD) and recovery (R) were calculated (Araujo, 2009) using the following formulas: 4CD=SDL¯100,{C_{\rm{D}}} = {{{S_D}} \over {\bar L}}100, where S_D is the standard deviation (SD) of the average lipid content in a series of 20 replicates [%] and L is the arithmetic average lipid content in a series of 20 replicates [%], 5LOD=b¯+3×SD blank ′{\rm{LOD}} = \overline {\rm{b}} + 3 \times {S_{D{\rm{ blank}}{{\rm{ }}^\prime }}} where b¯\overline {\rm{b}} is the arithmetic average lipid content in a series of 10 blanks [%] and S_{D blank} is the SD of the average lipid content in the blanks [%] and 6R=(Lw¯+C)−Lw¯T×100%,R = {{\left( {\overline {{L_{\rm{w}}}} + C} \right) - \overline {{L_{\rm{w}}}} } \over T} \times 100\% , where C is a weight or concentration of tripalmitate [mg, mg · cm⁻³ and μg · cm⁻³ for the gravimetric, Marsh and Weinstein method and Frings and Dunn method, respectively], L¯w{{\bar L}_{\rm{w}}} is an average weight or concentration of lipids in a series of 20 replicates [mg, mg · cm⁻³, μg · cm⁻³] and T is the known weight or concentration of tripalmitate added to the tube [mg, mg · cm⁻³, μg · cm⁻³].

Labour intensity (workload) of each lipid determination method was calculated as the average time (rounded to 1 min) needed for performing the analysis of one series of samples (n = 20), preparation of two series of standard solutions and standard stock solutions and, in case of the Frings and Dunn method, also preparation of the phospho-vanillin solution.

The Shapiro–Wilk normality test was used to verify the distribution of data, and the Levene test was employed to check the homogeneity of variance. The significance of differences between variables for the main effects and interactions between them was estimated by analysis of variance (ANOVA), followed by Tukey’s post-hoc test (α/n) when F was significant. The level of significance for all tests was p < 0.05. All statistical analyses were performed using Statistica 13.1 software (TIBCO Software Inc., San Ramon, USA).

Soft tissue lipid content varied significantly amongst the invertebrate species and three analytical methods for M. trossulus and C. crangon (ANOVA; Tables 1 and 2). The lipid content in the tissues of H. diversicolor did not differ statistically amongst analytical methods. Tukey’s post-hoc test (α/n) detected differences in the lipid content of mussel tissues between the gravimetric and Frings and Dunn methods (p = 0.036) and between paired species for all methods (p < 0.001).

The highest lipid contents in the mussels (mean ± standard error [SE]; 14.85 ± 2.95%, n = 20) and polychaetes (10.81 ± 2.36%, n = 20) were assayed with the gravimetric method and the lowest using the Frings and Dunn method (13.27 ± 1.20%, n = 20 and 10.64 ± 1.08%, n = 20, for M. trossulus and H. diversicolor, respectively; Table 1). The lowest lipid content in crustacean tissues (6.37 ± 2.08%, n = 18) was determined by the gravimetric method. The intermediate lipid contents in the invertebrate tissues, except C. crangon, were measured using the Marsh and Weinstein method. The lowest values of SE were recorded for the Marsh and Weinstein method, intermediate values for the Frings and Dunn method and the highest using the gravimetric method (Table 1).

The largest CDs in routine replicated assays for all analysed species were recorded for the data obtained using the gravimetric method, the lower CDs were calculated using the Frings and Dunn method, and the highest precision was obtained for the Marsh and Weinstein method (Table 3). The results for lipid content in the tissues of M. trossulus were the most accurate when using all analytical methods. For the gravimetric and Frings and Dunn methods, the least accurate results were obtained for lipid content in the crustacean tissues, whereas for the Marsh and Weinstein method, in the nereid tissues.

The lowest LOD for all species was obtained using the Marsh and Weinstein method, and the highest, except for the nereid tissues, using the gravimetric method (Table 3). The Frings and Dunn method provided intermediate LOD; exceptions were the nereid tissues for which the LOD value was highest.

Recovery ranged from 89.1% to 102.5% with the highest value obtained by the Marsh and Weinstein method, except for C. crangon tissues, for which R was 95.2% (Table 3). The lowest efficiency was described for the gravimetric method, except for C. crangon tissues, for which R was 101.7%.

An assay of 20 samples using the gravimetric method took only about 45 min, whereas analysis of a series of 20 replicates consumed 1 hr 15 min and 2 hr when using the Marsh and Weinstein and Frings and Dunn method, respectively (Table 4). In addition, the latter two procedures required prior preparation of SSSs, which ran for 20 min and 35 min, respectively. The preparation of the phospho-vanillin solution by the Frings and Dunn method took an additional 20 min. The average total time needed to complete the analysis using the gravimetric, Marsh and Weinstein and Frings and Dunn methods was estimated at 45 min, 95 min and 175 min, respectively.

Bulk lipid content in soft tissues of marine macrobenthic fauna varies with environmental conditions (e.g. temperature, salinity, food quantity and its nutritional value) and biological factors (e.g. species, physiological state, age and sex) and increases before reproduction and during periods of low food availability (Lehtonen, 2004; Wiklund et al., 2009). For example, lipid content in tissues of the brown shrimp C. crangon collected along the coast of Great Britain between late April and early May (i.e. before the reproductive period) ranged from 18% to 20% of dry weight (Moore, 1976). According to Szaniawska (1983), the lipid level in tissues of the same species from the Gulf of Gdańsk (southern Baltic Sea) fell in 2.6%–13.7% depending on the season whilst Mika et al. (2012) showed that in spring total lipid content of the shrimp muscle tissue from the Baltic Sea was only 3.4%. The results obtained in this study (6.4%–7.4%) are within the range of values presented by Szaniawska (1983) but differ from the data of Mika et al. (2012), which results likely from different sample types analysed i.e. muscle tissue versus whole organism.

In the shallow coastal waters of the southern Baltic, lipid content in soft tissue of the mussel M. trossulus varies seasonally, increasing from March to April, when it reaches its maximum level (24%), remaining high even until June (Szaniawska, 1983; Wołowicz et al., 2006). The accumulation of lipid reserves in the bivalves in this period is induced by elevated food consumption after spring algal blooms to recover after winter starvation and to prepare for reproduction. Lipid content measured in this study is fairly low (13.3%–14.9%), which probably stems from the earlier reproduction of the bivalves in the year of sampling. The release of lipid-rich gametes into the external environment induces a substantial reduction in total lipid content in the mussels. According to Martínez-Pita et al. (2012), during the pre-reproduction period, females of Mytilus galloprovincialis may contain up to twice as much fat as males. However, its reproductive effort might be so high that it can lead to a decrease in the physiological state and, in the extreme case, to the death of an organism (Wołowicz et al., 2006). Another reason for such a low lipid content in the mussels assayed in this study is the salinity variations in the overlying bottom zone in the shallow area where the mussels were collected (Sokołowski et al., 2017). In a laboratory experiment described by Lasota et al. (2018), variations in salinity resulted in a significant decrease in bulk lipid content in the soft-shell clam Mya arenaria and the Baltic clam Macoma balthica from the southern Baltic Sea. Salinity-induced osmotic shock can cause cell breakdown and lipid peroxidation (Abele & Puntarulo, 2004) or a change in the lipid-protein ratio (Sokołowski et al., 2003).

Empirical data on the lipid content in tissues of H. diversicolor from the Baltic Sea are scarce. The nereids from the coast of Great Britain have been reported to store 31.5% and 14% lipid reserves in winter and summer, respectively (García-Alonso et al., 2008). According to Wang et al. (2019), the bulk lipid content of H. diversicolor from the Norwegian coast ranged from 13% to 16% depending on its diet: similar lipid levels were noted in individuals with a diet based on microalgae and fish bits whilst lower values were observed for individuals fed with waste from juvenile salmonids and a mixture of fish waste and microalgae. Anglade et al. (2023) reported that the bulk lipid content was between 12.3% and 19.6% of the polychaetes reared on different types of aquaculture sludge in the Trondheimsfjord in Buvika, Norway. In this study, the lipid content was lower than that in the nereids from the Norwegian coast and ranged from 10.4% to 10.8%.

The quality of the results of biochemical analyses is affected by several factors e.g. diligence of sampling, sample pretreatment and laboratory procedures, analytical equipment and quality of reagents. In order to manage the quality of the analytical method, control analysis and analysis of procedural blank samples are routinely performed (Skoog et al., 2013). In this study, standard validation parameters were calculated for each method, including precision, LOD and recovery. Tukey’s post-hoc tests showed no significant differences in the bulk lipid content of a given species between paired analytical methods; the exception was M. trossulus, which demonstrated a significant difference between the results obtained using the gravimetric and Frings and Dunn methods. The observed differences were likely caused by a random experimental error resulting from e.g. accidental sample contamination, loss of a part of the weighed sample, instability of analytical equipment or analyst fatigue (Taylor, 2022). An increasing number of replicates usually offers a compromise leading to a reduction in random error but not to its total elimination. The lack of significant differences between the methods tested indicates that each method can be used successfully in analyses of both low-lipid and lipid-rich tissues of marine macrobenthic invertebrates. The calculated SD and R of each method fall in most cases in a satisfactory range of 0%–8% and 90%–100%, respectively (Taverniers et al., 2004). All methods used in this study proved to provide accurate and highly reproducible quantitative measures of total lipid content in the whole soft body of benthic invertebrates.

The Marsh and Weinstein method offers the most accurate analytical protocol for lipid quantification (the highest precision, low LOD, SD and SE) regardless of the analysed animal tissue. The obtained recovery (mean for three invertebrate species, 99.5%) was similar to that reported by Reisenbichler and Bailey (1991) (95%–100% depending on the analysed standard), who indicated that the possible difference in R amongst different standards might have been caused by the modification of the lipid extraction method. When considering all validation parameters, the best overall performance of the analyses was documented for the Marsh and Weinstein method, moderate for the Frings and Dunn method and worst for the gravimetric method. In addition, the recovery of the Frings and Dunn method fell into a wider range than that obtained using the Marsh and Weinstein method, but was narrower than that calculated for the gravimetric method. The exception was lipid analyses of crustacean tissues for which R varied from 94.9% to 101.7%. In the original study of Frings and Dunn (1970), precision in repeated assays of 60 randomly selected serum samples was 3.5% and Jacobs and Henry (1962) reported a CD of 9.6% for the gravimetric method. In terms of validation parameters, the Marsh and Weinstein method can be, therefore, recommended for quantitative total lipid determination in macrobenthic invertebrates of various fat content.

Besides high precision and accuracy, consideration should also be given to another important design criterion of the quantitative lipid determination method: workload. Estimates of the workload required to complete laboratory analysis can vary widely with analyst experience and practical skills, ease of operation of analytical equipment, organisation of laboratory workspace and complexity of the analytical procedure. According to Harvey (2000), depending on the technical advancement of the analytical equipment, the time spent on staff training should also be considered working time. Several steps in analytical procedures have, however, fixed times (e.g. centrifuging, drying, etc.), which do not vary with analyst or laboratory conditions. In the case of a single run of many series of samples, the most optimal in terms of time consumption are continuous and/or automated methods (Webster et al., 2005). However, these methods were not used in this study due to the small number of samples (n = 18–20).

The results of labour intensity estimates showed that bulk lipid content determination using the gravimetric method is the most time-effective. Two other methods were more complex and included several additional phases such as dosing reagents, sample incubation and optical density measurements, which all extended the total time required to complete the analysis. Larger quantities of the SSSs (and the phospho-vanillin solution in the case of the Frings and Dunn method) can be, however, used in subsequent analyses to reduce the average time for one series of samples. González et al. (2004) indicated that maintaining a balance between time consumption and reliability of the obtained results can be done by selecting the appropriate number of quality control analyses and choosing less or more accurate analytical methods. The gravimetric method is a fast alternative to colourimetric methods but has the lowest precision and the highest LOD values. The choice of analytical method requires, therefore, careful consideration of different criteria.