1 INTRODUCTION

Plankton, the basic component of the marine ecosystem, plays an important role in primary production and biogeochemical cycle (Taylor and Landry, 2018). Many of these processes are known to be size-dependent (e.g., Platt and Denman, 1977). Analyses of plankton size structure can help our understanding of not only the key property of the marine ecosystem, but also of the physiological rates and ecological functions (Marañón, 2015; Dong et al., 2018; Li et al., 2018; Dong et al., 2021; Liu et al., 2021a). As an empirical approach to studying biomass distribution through a function of planktonic organism size, the normalized biomass size spectrum (NBSS, Platt and Denman, 1978) has been widely used to describe planktonic size structure (Rodriguez et al., 1987; Quintana et al., 2002; García-Muñoz et al., 2014). Generally, the slope (b) of NBSS could represent the distribution of plankton biomass and its transfer efficiency to larger organisms (Blanco et al., 1994; White et al., 2008) with the intercept (a) of NBSS indicating the global property of marine ecosystems, such as total biomass or the total number of organisms (Blanco et al., 1994).

Despite the fast development of automated techniques for image analyses, a direct plankton assessment can still be challenging, particularly in rough seas. Sample preservation is thus a common step before microscopic analyses (Choi and Stoecker, 1989; Ohman and Snyder, 1991; Zinabu and Bott, 2000; Jaspers and Carstensen, 2009). Lugol's solution is widely used to fix plankton samples at sea (Ohman and Snyder, 1991; Montagnes et al., 1994; Martin et al., 2006; Liu et al., 2021b). Compared to the neutral Lugol's solution, the acid Lugol's solution could have a higher efficacy on preserving microflagellates (Hällfors et al., 1979; Jaspers and Carstensen, 2009; Williams et al., 2016) despite a higher risk of cell loss and shrinkage (Ngando and Groliere, 1991; Mukherjee et al., 2014). It is well known that a higher concentration of Lugol's solution may increase the cell breakage rate (Stoecker et al., 1994; Hawkins et al., 2005). Also, its impact on plankton can be species-specific with some species showing no significant effect but others with a large change in cell biovolume (Yang et al., 2017). However, the Lugol's effect on the plankton community as a whole remains less addressed as previous studies were mostly focused on its fixation efficiency on individual taxa (Zarauz and Irigoien, 2008).

Sample storage time has been suggested as an important factor affecting the accuracy of plankton cell-biovolume and abundance estimations (Zinabu and Bott, 2000; Menden-Deuer et al., 2001). A range of storage times for Lugol-preserved plankton samples has been used in the literature, such as 24 h (Jakobsen and Carstensen, 2011), 65 h (Ventura and Jeppesen, 2009), three weeks (Mukherjee et al., 2014), and four months (Zarauz and Irigoien, 2008). Some studies did not report the time of sample analysis, assuming no temporal change (Huete-Ortega et al., 2010; Moreno-Ostos et al., 2015). It is still unknown how the fixed plankton community varies through time during the sample storage. Therefore, it is desirable to assess the regulating factors for temporal variation of the Lugol-fixed community, which should be critical not only for accurate estimation of plankton community but also for comparative analyses of different datasets.

In this study, a broad size range of plankton organisms with equivalent spherical diameters (ESD) of 10–200 μm was analyzed for the surface samples collected at a coastal station of the northern South China Sea (NSCS). The plankton size information was obtained by a Flow Cytometer and Microscope (FlowCAM) system (Sieracki et al., 1998) following the previous methods for individual taxa and natural communities (Zarauz and Irigoien 2008; García-Muñoz et al., 2014). A comparison between the fresh and the Lugol-fixed samples was conducted to assess the preservative efficacy of the acid Lugol's solution. In addition, temporal change in the Lugol-fixed plankton community was investigated by regular sampling over a short-term (a week or a month) or a long-term (nearly seven months) period of storage time. Statistical analyses were used to assess the evolution of the Lugol-fixed community (in terms of the NBSS parameters). Finally, we discussed the potential factors controlling the observed temporal patterns.

2 MATERIAL AND METHOD 2.1 Field sampling and experimental design

Field samplings were conducted at a station near Wanshan Island (113.8°E, 21.9°N) outside the Zhujiang (Pearl) River estuary of the northern South China Sea in December 2018, May 2019, and June 2019 (Fig. 1). During each sampling date, a surface seawater sample (~2 m below the sea surface) was collected by a clean bucket, and then passed through a 200-μm mesh to remove large grazers. The sample was then transferred to several clean plastic water bags and stored in the dark for less than 3 h before the laboratory analyses. After fixing with a 1% acid Lugol's solution, the samples were stored in a dark container at room temperature throughout the entire period of experiments. The acid Lugol's solution was prepared by dissolving 100 g of potassium iodide and 50 g of iodine in 1 L of 10% glacial acetic acid (Ohman and Snyder, 1991). In each experiment, both the fresh and the fixed samples were analyzed in triplicate for the plankton community using a FlowCAM method. Based on the duration of sample storage, three independent experiments were performed to explore the temporal variations of the Lugol-fixed community: 1) one-week experiment (Exp. 1) starting from May 27, 2019; 2) one-month experiment (Exp. 2) starting from June 10, 2019; 3) a long-term experiment (Exp. 3) nearly seven months starting from December 18, 2018.

2.2 FlowCAM method

A portable dynamic imaging particle size analyzer (Flow Cytometer and Microscope, FlowCAM Portable Model) was used to obtain the images of plankton and the data of plankton size. It has the advantage of identifying particles (such as debris) that cannot be quantified by the classic methods (Le Bourg et al., 2015). The efficiency of the FlowCAM method has been well documented by comparing with results from the microscopic measurements (Ide et al., 2008; Álvarez et al., 2011).

Each time, a 40-mL sample was analyzed in AutoImage mode with sample fluid was pumped through an imaging system via a flow cell (FC200) using a C71 model Syring Pump at a constant speed of 8 mL/min under 40× magnification. The detection limit was 5 μm of ESD. Each sample was gently stirred with a bar to prevent particle sedimentation before being pumped by a syringe pump (Cisternas-Novoa et al., 2015). All particles of 10–200-μm ESD were photographed, but some invalid pictures of bubbles and repetitive pictures were excluded (Zarauz and Irigoien, 2008). Plankton fragments and other organic particles were kept in the database as they could be important for the carbon cycle according to Roy et al. (2000). The volume of each particle was automatically calculated by FlowCAM according to its ESD. To account for the cell shrinkage caused by fixation, the volume of particles in the preserved samples was corrected by a factor of 1.33 (Montagnes et al., 1994): Volume_{live cells}=1.33×Volume_{fixed cells}. Three different size groups were classified by FlowCAM particles: 10–20 μm (nanoplankton), 20–40 μm (small microplankton), and 40–200 μm (large microplankton). The total particle abundance for each sample within each size range is automatically calculated once all the particles have been identified.

2.3 Normalized biomass size spectra

A method of normalized biomass size spectra (NBSS) was used for plankton analysis in this study. The normalized biomass was constructed by dividing the total biovolume (biovolume of each particle was calculated based on its ESD) of each size class by the width of the size class (Platt and Denman, 1978). The minimal volume of each size class was chosen as a nominal size (Blanco et al., 1994).

The calculation of NBSS can be represented by the following equation:

where A_i is the total volume of particles in each size class and ∆V_i represents the width of each size class. The ∆V_i is calculated according to the formula: ∆V_i=V_i+1–V_i, where V_i is the minimal volume in each size class. Particles with sizes ranging from 10 μm to 200 μm were arranged into 13 groups without considering the taxonomy. The series of size classes was established following a geometric 2n series approach (Sheldon et al., 1972), which can be represented by V_i+1=2V_i with V₁=5.23×10² μm³. Parameter a and b are constants, which can be calculated through linear regression. For the parameters of NBSS, the slope (b) can reflect the size distribution in different size classes while the intercept (a) could represent the plankton community abundance and the productivity of the ecosystem (Blanco et al., 1994; White et al., 2008).

2.4 Statistical analysis

The figures were constructed using Microsoft Excel 2010 and Origin Pro 2018. Statistical analyses of linear regression and Pearson correlation analysis were performed using software SPSS v.22.0 (SPSS). A permutation test was carried out to determine the significance of the regression and to calculate the Pearson correlation coefficients with R representing the Pearson coefficient of correlation. Results were considered significant at a P < 0.05 and strongly significant at a P < 0.01.

3 RESULT 3.1 Temporal change of the NBSS parameters

Generally, the linear fit of NBSS was significant for all subsamples (R > 0.9, P < 0.01). For the non-preserved sample, the slope (b) of NBSS varied from -0.89 to -0.57, which should reflect a large difference in the raw seawater samples among the three experiments. The NBSS slope of -0.57 on May 27, 2019 was not much different from -0.65 on June 10, 2019, but much higher than -0.89 on December 18, 2018. The change of NBSS slope over seasons may indicate a seasonal change of phytoplankton community structure. Besides, a large difference in the NBSS intercept (from ~12.85 in May 2019, to ~15.38 in June 2019, and to ~17.97 in December 2018) may reflect the seasonal variability of particle abundance with higher intercept for higher total particle abundance.

Overall, the NBSS slopes of the Lugol-fixed samples were lower than the fresh samples, although there were generally higher NBSS intercepts in the preserved samples. Linear changes of slope and intercept with storage time were generally evidenced in the short-term experiments. For the one-week experiment (Fig. 2a), we found a decreasing slope from -0.57 to -0.69 at a rate of -0.021/d (R=0.93, n=24, P < 0.01) and an increasing intercept from 12.85 to 15.25 at a rate of 0.099/d (R=0.96, n=24, P < 0.01). Meanwhile, there was a lower rate of decrease in slope (-0.006/d, R=0.89, n=54, P < 0.01), together with a lower rate of increase in intercept (0.026/d, R=0.87, n=54, P < 0.01), as evidenced in the one-month experiment (Fig. 2b). Our results reveal a linear increase of the fraction of small-sized plankton but a linear decrease of the total number of organisms with storage time.

For the long-term experiment, the response of the NBSS parameters with time can be separated into two different phases with a turning point occurring three months after the preservation (Fig. 2c). In the first phase, the slope decreased linearly from -0.89 to -1.18 at a rate of -0.002/d (R=0.88, n=12, P < 0.01) with a linear increase of intercept from 17.97 to 23.31 at a rate of 0.013/d (R=0.84, n=12, P < 0.01). The changing rates of slope and intercept during the first phase are comparable to those found in the one-month experiment. However, the NBSS slope started to increase from -1.18 to -1.05 at a rate of 0.001/d (R=0.77, n=15, P < 0.05) during the second phase after three months of preservation, accompanied by a slight decrease of intercept from 23.31 to 20.76 with a rate of -0.006/d (R=0.90, n=15, P < 0.01). The opposite trends of slope and intercept between the first and the second stages should indicate a fundamental change in plankton particles and size structure over long-term preservation. There was an increase of the fraction of large-sized plankton but a decrease of the total plankton abundance in the long-term preservation compared to those of the short-term preservations.

3.2 Changes of NBSS spectrum during the long-term experiment

To explore the significant size structure shift found in the long-term experiment, we further examine the detail of the NBSS curve. The plankton size spectrum varied substantially from the initial condition to the turning point of three months later, and to the very end of seven months later (Fig. 3).

Compared to that of the initial fixed sample (Fig. 3), the NBSS curve of the sample after 96 d of preservation was much flatter in the range of 10 to 20 μm, but steeper in the range of 40 to 200 μm (Fig. 3). The spectrum changed in the size range of 10–20 μm should indicate an increase in particle abundance and the more dominance of large particles. At the same time, for particles within the range of 40–200 μm, there was a reduced abundance of larger particles but an increased fraction of smaller particles. In the second stage, however, the NBSS curve of the sample with 208 d of preservation was steeper in the range of 10–20 μm but flatter in the range of 40–200 μm than that of the sample after 96 d (Fig. 3). These changes in the NBSS curve might imply that the particle size structure had undergone a different way of change in the second stage. After three months of preservation from the 96^th day to the 208^th day, a steeper NBSS curve in the size range of 10–20 μm should indicate that an increased fraction and abundance of smaller-sized particles. However, the trend of the NBSS curve became flatter in the size range of 40–200 μm (Fig. 3) during the second stage, which should reflect an increase of fraction but a reduced abundance of large-sized particles.

3.3 Examples of the visible size changes of planktonic particles

The dominant components of our plankton samples collected from the Wanshan station were diatoms and ciliates according to the particle images obtained by FlowCAM (Fig. 4). At the beginning of our experiment (t=0), we can identify the cell structures of planktonic particles with different size ranges (Fig. 4a & e). During the first phase of the long-term experiment, there was an increased abundance of diatoms with shorter chains in the size range of 10–40 μm for the fixed sample (t=96) (Fig. 4h) compared to that of the initial sample (Fig. 4e). Also, we could still identify some spindle-shaped diatoms even after 96 d of preservation (Fig. 4h). Comparing the difference in particle images between the initial sample and the sample of the 96^th day (Fig. 4e–h), we could find visible disintegration of the chain-forming diatoms and fragmentation of ciliates during this period of time. These findings may explain the decrease of the NBSS slope found during the first phase of the long-term experiment (Fig. 2c). The phenomenon of chain-forming diatoms broken into pieces was also well documented in our short-term experiment starting on June 10, 2019 (Fig. 4a–d). However, there were increasing particle aggregates but decreasing chain-forming diatoms in the size range of 20–200 μm from the 96^th day to the 208^th day (Fig. 4i–l). Therefore, there was an intense particle accumulation during this period, which had resulted in a slight increase of NBSS slope but a moderate decrease of NBSS intercept in the second phase of the long-term experiment.

3.4 Temporal changes in biovolume and abundance of different plankton size-classes

Generally, the total biovolume and the total abundance of particles in the preserved sample varied substantially with the storage time. Also, there was a large difference in the temporal patterns of particle biovolume and abundance among three different size classes, including nanoplankton (10–20 μm), small microplankton (20–40 μm), and large microplankton (40–200 μm).

We did not find any significant correlations between biovolume and storage time in the short-term experiments (Fig. 5a–b). However, the total biovolume showed a linear increase and then a linear decrease with the storage time during the long-term experiment (R > 0.7, P < 0.01, Fig. 5c). Furthermore, the fraction of nanoplankton biovolume increased linearly with the storage time during the first stage of the long-term experiment (R=0.72, n=12, P < 0.01, Fig. 5c), which was well consistent with those found in the two short-term experiments (R > 0.7, P < 0.01, Fig. 5a–b). Our results suggested that the preserved community would evolve towards a smaller size range during the first stage of preservation within three months. Interestingly, the relationship between biovolume fraction and storage time was opposite for small and large microplankton during the second phase (from 96^th to 208^th day). A linear decrease of biovolume with storage time for small microplankton but a linear increase for large microplankton was found when the plankton sample was preserved for over 96 d (R > 0.6, P < 0.01, Fig. 5c). These results revealed that the efficiency of acid Lugol's solution on sample preservation could vary among different plankton size classes.

The total particle abundance was found to increase linearly with storage time during the one-week experiment (R=0.69, n=24, P < 0.01; Fig. 6a), the one-month experiment (R=0.61, n=54, P < 0.01; Fig. 6b), and the first phase of the long-term experiment (R=0.86, n=12, P < 0.01; Fig. 6c). However, the total abundance decreased linearly with storage time during the second phase of the long-term experiment after 96 d of preservation (R=-0.83, n=15, P < 0.01; Fig. 6c). We found a large reduction in the fixative effect of Lugol on plankton abundance after three months of storage leading to an increasing plankton loss. Generally, there was no significant correlation between nanoplankton fraction (abundance) and storage time, except for the second phase of the long-term experiment (R=0.95, n=15, P < 0.01; Fig. 6c). For the long-term experiment, the fraction of small microplankton linearly increased with time (R=0.65, n=12; P < 0.05) within three months but linearly decreased with time (R=-0.95, n=15; P < 0.01) after three months of preservation. We found a decreasing role of 20–40 μm sized micro-particles in the second stage of preservation with some micro-particles broken up and contributing to an increase of nano-particles. Interestingly, there was a linear decrease of large microplankton fraction with time during both of the short-term experiments (R > 0.4, P < 0.05; Fig. 6a–b) and the first phase of the long-term experiment (R=0.7, n=12, P < 0.01; Fig. 6c), which should indicate a significant loss of large cells during short-term preservations by the Lugol's fixation.

4 DISCUSSION

It has been known for some time that Lugol's fixation may cause cell shrinkage based on experimental studies (Menden-Deuer et al., 2001; Martin et al., 2006; Álvarez et al., 2014). In addition, previous studies indicated that plankton aggregates and fragments induced by Lugol's solution could lead to substantial changes in plankton size structure (Zinabu and Bott, 2000; Zarauz and Irigoien, 2008; Mukherjee et al., 2014). It was speculated that the efficacy of Lugol's fixation on plankton preservation could vary with the storage time (Zarauz and Irigoien, 2008). A large difference in plankton size structure between fresh and preserved samples revealed by a microscopic study was attributed to a long duration of storage (Williams et al., 2016). Our results provide strong evidence for changes in the size structure of the Lugol-fixed plankton community with storage time. In particular, we found the temporal response of the plankton community to Lugol's fixation can be size-dependent. Likely, this was a result of the different sensitivity of various plankton species in response to Lugol's fixation (Menden-Deuer et al., 2001; Hawkins et al., 2005).

We found a decreasing NBSS slope but an increasing NBSS intercept with time for the Lugol-fixed community within three months of storage and these could be attributed to the decomposition of small-sized microplankton leading to an increase of the total particle abundance. Similar processes have been known to affect the size distribution of plankton in the field (Blanco et al., 1994). It was also known that sample preservation for plankton size structure could be affected by particle fragmentation (Zarauz and Irigoien, 2008). The particle fragments that mostly originated from broken cells and fecal materials could bias the estimation of the size spectrum of organic particles. We found that both the fraction of small-sized particles (10–40 μm) and the total particle abundance linearly increased with time within a short term of preservation. This result could be attributed to an increased intensity of particle fragmentation, as the acid Lugol's solution could readily break down the cellular structure of some organic particles. For example, it was suggested that the cellular structure of diatoms can be destroyed during preservation due to the lower pH of acid Lugol's solution (Hällfors et al., 1979). Moreover, the acidic Lugol's solution could result in a decrease of biovolume for protozoa (Choi and Stoecker, 1989). In addition to mechanical handling such as shaking and mixing during the sampling process, the Lugol's fixation could increase sample fragility leading to cell breakage and chain fragmentation. Indeed, we found increased concentrations of broken diatom chains and cell fragments in nanoplankton (10–20 μm) and small microplankton (20–40 μm) under the effect of acid Lugol's solution. Our results support the previous finding that acidic Lugol's solution can cause plankton cell breakage and chain fragmentation (Mukherjee et al., 2014). Our result is also consistent with the report of the increased fraction of small plankton after four months of preservation by Lugol's solution (Zarauz and Irigoien, 2008). Therefore, within a short-term preservation time (less than three months), large organic particles (40–200 μm) such as ciliates and chain-forming diatoms would break into smaller-sized pieces under the effect of particle fragmentation induced by the acid Lugol's solution. This would in turn lead to variation in the plankton community size structure by increasing total particle abundance and decreasing the fraction of large microplankton.

It is interesting to find that the NBSS parameters start to show opposite trends after reaching extreme values at about three months after preservation. Different from those in the short-term experiment (from initial to the 96^th day), the proportion of small micro-particles (20–40 μm) and total particle abundance generally decreased with time during the second phase of the long-term experiments (from the 96^th day to the 208^th day). These results indicated that the effect of Lugol's solution on the plankton community may be very different between short-term and long-term preservations. It has been suggested that Lugol's fixation could cause shrinking of Ceratium fusus and Scrippsiella trochoidea during the first month of preservation but swelling of these dinoflagellates after three months of preservation (Menden-Deuer et al., 2001). This was different from the continuous shrinking of some diatom species after eight months of preservation (Menden-Deuer et al., 2001). An increasing cell loss of planktonic ciliates with time was reported after nine months of preservation (Ngando and Groliere, 1991).

The opposite trends between short-term and long-term preservations could be explained by the interplay between particle aggregation and particle fragmentation. A continuous increase in the fraction of nanoplankton during our study should reflect a long-standing trend of the particulate matters being imported into the size range of nanoplankton due to the persistent fragmentation of larger-sized planktonic particles. However, the process of planktonic organisms broken into smaller particles would be substantially weakened after three months of preservation as evidenced by FlowCAM image analyses. Consequently, particle aggregation became more important than particle fragmentation during the long-term period of preservation leading to a decrease of small micro-particles (20–40 μm) but an increase of large particles. Our findings are consistent with the previous reports that particle aggregations induced by the Lugol's solution could lead to higher concentration of large aggregates and organic detritus and subsequently affect the size-spectrum of the plankton community (Zinabu and Bott, 2000; Zarauz and Irigoien, 2008).

5 CONCLUSION

In this study, laboratory experiments were conducted to examine the temporal change of the Lugol-fixed plankton community in terms of the NBSS parameters and the spectrum curves estimated by FlowCAM. The effect of Lugol's solution on the plankton community can generally be divided into two stages according to the storage time. In the first stage of short-term preservation (less than 3 months), the process of breaking large-sized particles into smaller size levels is more dominant. In the second stage of long-term preservation (over 3 months), the process of fragmentation is weakened, so the process of small-sized particles aggregated into larger size levels becomes relatively more dominant. The Lugol's fixation had changed the plankton size-structure and biomass in an unfortunate way, causing a loss of precision when Lugol-fixed materials are assessed. Therefore, it will be worthwhile performing direct field assessments using FlowCAM to avoid sample fixation. It is also important that the sample storage time should be taken into account when analyzing or comparing the data of Lugol-preserved samples.

Of course, more researches are needed to quantify the actual impacts of Lugol's fixation on the natural samples with different community compositions and productivities. This is because the plankton response to the Lugol's effect could vary substantially among different plankton species and size classes. Direct measurement in the field by FlowCAM with high efficiency and accuracy is highly recommended for studying the size structure of the plankton community to avoid the influence of sample fixation and preservation.

6 DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.

7 ACKNOWLEDGMENT

We thank Mr. Guanghui LI for help during the field sampling.

Electronic supplementary material

Supplementary material (Supplementary Fig.S1) is available in the online version of this article at https://doi.org/10.1007/s00343-021-1155-1.