The role of transparent exopolymer particles (TEP) in the increase in apparent particle stickiness (α) during the decline of a diatom bloom

Abstract

The termination of diatom spring blooms in temperate waters has been connected with the formation and subsequent rapid sedimentation of aggregates. According to coagulation theory, the rate of aggregate formation depends on the probability of particle collision and on the efficiency with which two particles adhere once they have collided (stickiness). During this study, the variation in particle stickiness was determined over the decline of a diatom bloom using the Couette Chamber assay with low shear (G = 0.86 s^–1). A mixed diatom population, dominated by Skeletonema costatum, was sampled during the spring bloom in the Baltic Sea and incubated in the laboratory for 18 days. Measurements of diatom species composition, transparent exopolymer particles (TEP) and bulk particle abundance, as well as chemical and biological variables, were conducted in order to reveal the determinants of coagulation efficiency. The investigation showed that an increase in TEP concentration relative to conventional particles at the decline of the bloom significantly enhanced apparent coagulation efficiencies. High proportions of TEP led to apparent values of stickiness >1, which indicates that collision rates can be substantially underestimated when the stickiness parameter α is calculated on the basis of conventional particle counting only, e.g. with the Coulter Counter. A new stickiness parameter, α′, was therefore estimated based on the combined volume fractions of TEP and conventional particles. The problems of stickiness measurements are discussed and the role of TEP in coagulation processes is emphasized.

Introduction

The role of large particle aggregates (marine snow) in the vertical flux of organic matter into the oceans' interior is widely acknowledged (Shanks and Trent, 1980; Fowler and Knauer, 1986; Asper et al., 1992; Gardner, 1997; Jackson and Burd, 1998). Theoretical analyses of particle coagulation processes predict that aggregate formation depends on the probability of particle collision and on the efficiency with which two particles that collide stick together afterwards (stickiness) (Hunt, 1982; McCave, 1984; Jackson, 1990). The former is a function of particle concentration, size and the mechanism by which particles are brought into contact, e.g. Brownian motion, shear or the differential settlement of particles. The latter depends mainly on the physicochemical properties of the particle surface and may vary with the particle type.

In the ocean, high particle concentrations occur during phytoplankton blooms and aggregates have readily been observed at these times (Smetacek, 1985; Alldredge and Gotschalk, 1989; Riebesell, 1991a, b; Tiselius and Kuylenstierna, 1996). However, the occurrence of aggregates does not always coincide with the peak of phytoplankton abundance. Rather, it is postponed towards the decline of the bloom (Smetacek, 1985; Riebesell, 1991a, b). This has been hypothesized to be due to an increase in particle stickiness (Smetacek, 1985), but has not so far been demonstrated empirically. Despite the importance of coagulation processes for biogeochemical cycles as well as for food web dynamics (Bochdansky and Herndl, 1992; Green and Dagg, 1997), information on the magnitude and variability of particle stickiness is relatively scarce. There are only a few direct measurements of phytoplankton stickiness (Kiørboe et al., 1990; Kiørboe and Hansen, 1993; Drapeau et al., 1994; Waite et al., 1997) and even less describe its variability during the course of natural phytoplankton blooms (Kiørboe et al., 1994, 1998; Dam and Drapeau, 1995).

Changes in particle coagulation efficiency have been attributed to the abundance of single species or as part of the life cycle strategy of cells (Smetacek, 1985; Kiørboe and Hansen, 1993; Crocker and Passow, 1995). Recently, a special class of particles was found to be readily abundant during phytoplankton blooms in water and aggregates as well. These transparent exopolymer particles (TEP) (Alldredge et al., 1993) are thought to play a central role in coagulation processes for two reasons. First, they are supposed to be very sticky and secondly their abundance may enhance the probability of particle collisions (Passow et al., 1994; Jackson, 1995). TEP are abundant in shelf seas (Passow and Alldredge, 1995; Schuster and Herndl, 1995; Mari and Burd, 1998) and in the open ocean (Engel et al., 1997; Hong et al., 1997). However, there is a lack of direct measurements that corroborate the role of TEP in the coagulation of particles.

In this study, the variability of particle coagulation efficiency was examined during the decline of a natural diatom bloom. The stickiness parameter (α) was determined using the Couette Chamber assay and related to the abundance of TEP, diatom species and Coulter Counter-detectable particles. The methodological problems of stickiness measurement are discussed in connection with the role of TEP in coagulation processes.

Method

Sample collection and incubation

Phytoplankton were sampled during a diatom spring bloom in the western Baltic Sea (Kiel Bight) on 24 March. A volume of 10 dm³ was sampled with Niskin bottles at 2 m depth. In the laboratory, the sample was filtered through a 200 μm mesh in order to remove the larger zooplankton and diluted with 14 dm³ of 0.2 μm (Nuclepore) pre-filtered sea water. Five aliquots of 4.5 dm³ were taken from the diluted sample and incubated in Plexiglas tubes. The tubes were well aerated by bubbling and received a photosynthetically active radiation flux of 210 μmolm^–2 s^–1, with a light:dark cycle of 12 h:12 h. The incubation was performed at a temperature of 15°C from 24 March to 11 April 1997.

Biological and chemical analysis

Analyses of biological and chemical variables were made daily from day 1 to day 6 and every second day thereafter. Aliquots of 100 cm³ were placed into polycarbonate bottles and frozen at –21°C for subsequent measurement of phosphate and nitrate with an autoanalyzer [after (Grasshof et al., 1983)]. Particulate organic carbon (POC) and nitrogen (PON) were determined with a Hereaus CHN analyzer (CHN-O-rapid) from duplicate 200 cm³ samples filtered ontopre-combusted glassfiber filters (Whatman GF/F). A total of 200 cm³ were filtered onto duplicate GF/F filters and chlorophyll (Chl) a was measured following the procedure of Jeffrey and Humphrey (Jeffrey and Humphrey, 1975). The concentration and size distribution of solid particles, 4 μm < equivalent spherical diameter (ESD) <60 μm, were determined with a Coulter Counter (Coulter Multisizer II) from three replicate 2 cm³ samples. TEP concentration was measured colorimetrically after Passow and Alldredge (Passow and Alldredge, 1995) and microscopically after Passow and Alldredge (Passow and Alldredge, 1994). For the colorimetric analysis, 100 cm³ were filled into polycarbonate bottles and preserved with formalin (1% final concentration). For analyses, three replicates of 20 cm³ each were filtered onto 0.4 μm polycarbonate filters (Nuclepore). Semi-permanent TEP slides were prepared in duplicates from 5 cm³ within 1 h after sampling. For analyses, slides were transferred to a compound light microscope and screened by a Sony color video camera on Super VHS with ×200 magnification. About 25–40 frames per sample were chosen randomly and digitized on a Macintosh PPC with an optical resolution of 2.71 μm³ per pixel. TEP were enumerated and sized semi-automatically by the image analysis program NIH Image 6.1 ppc., a public domain program developed at the US National Institutes of Health. The ESD and the equivalent spherical volume (ESV) of individual TEP were calculated from area measurements assuming the symmetry of a sphere (Mari and Kiørboe, 1996). Only TEP that did not touch the edge of the frame and contained at least 20 pixels, equivalent to a minimum TEP size of 8 μm, were counted.

Diatoms were identified and enumerated from 10–25 cm³ Lugol-fixed samples using an inverted microscope under ×400 magnification (Utermöhl, 1958).

Determination of particle stickiness

A value for the efficiency of particle coagulation is the probability that two particles which collide stick together:

stickiness (α) = adhesion rate/collision rate

with values ranging from 0 to 1. The parameter α has been determined empirically with experimental systems that provide information about particle collision rates, e.g. Couette Chambers (Kiørboe et al., 1990; Kiørboe and Hansen, 1993; Drapeau et al., 1994). In a Couette Chamber, the sample is incubated within a small gap between two cylinders. Rotation of the outer cylinder produces laminar shear of the fluid and leads to collision of suspended particles. Mean shear (G_m; s^–1) in this system can be calculated according to van Duuren (van Duuren, 1968):

\[\mathit{G}_{m}\ =\ 4\ {\pi}\ {\omega}\ \mathit{r}_{2}\ \mathit{r}_{1}\ (\mathit{r}_{2}^{2}\ {\mbox{--}}\ \mathit{r}_{1}^{2})^{{\mbox{--}}1}\ (1)\]

where r₂ and r₁ are the radii (cm) of the outer and the inner cylinder, respectively, and ω is the angular velocity (s^–1). Aggregation reduces the numerical abundance of particles within the chamber according to the equation:

\[\mathit{C}\mathit{_{t}}\ =\ \mathit{C}_{0}\ exp^{{\mbox{--}}\left(7.82\ {\alpha}\ {\phi}\ \mathit{G}_{m}/{\pi}\right)\mathit{t}}\ (2)\]

(Kiørboe et al., 1990) where ϕ is the volume concentration of particles (p.p.m.) and C_t and C₀ are the particle concentrations (no. dm^–3) at times t and 0, respectively. The stickiness can then be calculated by solving equation (2) for α. Although the calculation of α is based upon the assumption that all particles are approximately equal in size and stickiness, natural samples of course comprise a plethora of different types of particles usually distributed continuously along the size spectrum. Therefore, particle concentration is considered here as C_l (no. cm^–3), the number of particles that are determined with the Coulter Counter when using the 100 μm aperture, equivalent to the size range 4–60 μm ESD. The stickiness parameter, α, is then considered as the apparent stickiness.

The Couette Chamber used in this study is similar in design to that of Drapeau et al. (Drapeau et al., 1994). It consists of a horizontal chamber with a total length of 30 cm and radii of 5.75 and 4.5 cm for the outer and inner cylinders, respectively. A movable end cap that rotates together with the outer cylinder enables the displacement of the fluid as the sample is let out through a 0.5 cm tube at the opposite side. For each measurement of α, the Couette Chamber was filled with 1 dm³ of sample and run at a shear rate of G = 0.86 s^–1. About 50 cm³ were removed carefully after t = 1, 60, 120 and 180 min of incubation without stopping tank rotation. The numerical concentration and size distribution of particles were measured immediately after sampling with the Coulter Counter as described above. Dilution of the sample was not necessary as coincidence of particles at the aperture remained <5%.

Results

Development of chemical and biological constituents during the bloom

Nutrient concentrations decreased sharply within the first 4 days of the study. PO₄ decreased from 0.1 to 0.04 μmol dm³ and varied around 0.03 μmol dm³ afterwards. NO₃ started at 0.76 μmol dm^–3, but was not detectable at day 4 and day 5 (Figure 1a). During the rest of the incubation, NO₃ varied between the detection limit (0.01 μmol dm^–3) and 0.31 μmol dm^–3. The initial Chl a concentration was 6.6μg dm^–3 and declined gradually to a value of 1.5 μg dm^–3 until the end of the investigation (Figure 1b). Total abundance of diatoms started at roughly 10⁴ cells cm^–3 and decreased to 2.5 × 10³ cells cm^–3 on day 18 (Figure 1c). After day 5, the abundance of flagellates increased, but was not counted. Species composition was dominated by Skeletonema costatum with Chaetoceros species such as C.decipiens, C.borealis, C.gracile and C.curvisetum also present.

Initial POC concentration had doubled by day 4 to 494 μg dm^–3 and then declined to a minimum value on day 10 (Figure 2a). In contrast to Chl a, the POC content increased again after day 10 and reached a maximum value at day 18. PON was closely related to POC, but POC:PON exceeded the Redfield ratio of 6.6 (Redfield et al., 1963) during the whole study.

TEP concentrations fluctuated around 1300 μg xanthan equivalent dm^–3 until day 8, increased gradually until day 16 and sharply thereafter to reach a maximum value of 5600 μg xanthan equivalent dm^–3 on day 18 (Figure 2b). Xanthan equivalent values were significantly related to total area of TEP determined by image analysis (P < 0.001; data not shown), showing good comparability of both methods.

During the whole study, the numerical concentration of Coulter Counter-detectable particles (CCP) >4 μm ESD ranged between 1.0 × 10⁴ and 2.5 × 10⁴ cm^–3. In comparison, the abundance of TEP >8 μm ESD was about two orders of magnitude lower during the first part of the study (Figure 3a and b). Thereafter, TEP abundance increased clearly and reached a maximum of 5.25 × 10² cm^–3 on day 18. Since the mean size of TEP (~16 μm ESD) was larger than that of CCP (~6 μm ESD), the volume concentration of CCP was approximately only three times the volume concentration of TEP at the beginning and roughly the same at the end of the study (Figure 3c and d). Highest ratios of TEP:CCP by volume were observed on day 14 and day 18, respectively.

Variability of particle coagulation efficiency

The apparent stickiness (α) of particles was <0.1 during the first 4 days of the study. From day 5 to day 8, moderate values of α were determined, fluctuating around 0.2, then they increased greatly and even exceeded a value of 1 on days 14 and 18 (Figure 4).

TEP are not detected by the Coulter Counter [(Alldredge et al., 1993); personal experience], but are most likely involved in the coagulation process. As a first step to account for higher particle collision rates in the presence of TEP, α was calculated again with the initial volume fraction of TEP added to the initial volume fraction of CCP (denoted as α′ from now on).

The difference between values of α′ and α was small for day 2 to day 8 when the TEP fraction was low. However, on days 14 and 18 when TEP:CCP was highest, the inclusion of TEP volume concentration reduced α′ to values of <1 (Figure 4).

As the apparent stickiness increased during the decline of the diatom bloom, values of α were negatively related to the Chl a concentration (P < 0.005; Figure 5), diatom abundance (P < 0.05; Table I) and CCP concentration (P < 0.001). Some diatom species are supposed to be sticky by themselves (Kiørboe and Hansen, 1993; Crocker and Passow, 1995). In order to assess whether the represented species were responsible for coagulation efficiency during this study or not, the abundance of single species was correlated with the apparent stickiness (Table I). In no case were the two positively correlated; where the correlation was significant, it had a negative value, following the general decline in cell abundance during the experiment.

A significant positive relationship to the apparent particle stickiness was attained with TEP (P < 0.01), which was even stronger when the relative portion of TEP was considered, either as the ratio of numerical concentrations, i.e. TEP:CCP (P < 0.001), or as bulk variable ratio, i.e. the colorimetrically determined TEP concentration to the Chl a concentration (P < 0.001) (Figure 6). This supported the results of Dam and Drapeau, who observed that a decrease in α during a diatom bloom was related to a decrease in the chlorophyll-specific TEP concentration (Dam and Drapeau, 1995). As also shown in Dam and Drapeau, the line of the linear regression of α versus Chl a-specific TEP concentration intersects the ordinate at a negative value, suggesting that coagulation of particles needs a minimum amount of TEP (Dam and Drapeau, 1995).

Discussion

Methodological problems in the assessment of particle stickiness

For the investigation of particle coagulation processes, the Couette Chamber has been appreciated as an experimental system that not only provides a laminar flow field, but also allows quantification of shear by simple equations. In order to measure coagulation efficiency, the decrease in particle concentration within the chamber has to be determined. Recently, Waite et al. suggested that subsampling of the chamber and electronic particle counting may destroy aggregates, which would in turn lead to an underestimation of α (Waite et al., 1997). However, during this study, high apparent stickiness of particles was determined with values of α >1. Since α is taken as the probability of adhesion of two particles, values >1 obviously indicate an overestimation of α. To solve this problem, it has to be recalled that the conventional calculation of α is based on particles that can be enumerated with electronic counting systems. TEP are thought to play a substantial role in coagulation processes, but cannot be detected quantitatively with the Coulter Counter [(Alldredge et al., 1993); personal experience]. As a consequence, particle collision rates within the Couette Chamber might be severely underestimated mainly for two reasons. First, TEP increase particle concentration and, secondly, after sticking has succeeded they enlarge the size of CCP. During this investigation, values ofα > 1 were found when the ratio of TEP:CCP was high and exceeded a value of 2 as the volume concentration of TEP was twice the volume concentration of CCP. Addition of the TEP to the initial volume concentration of particles resulted in a decrease in the new stickiness parameter α′ to values of <1. This shows that a part of the supposed ‘stickiness' can be explained just by elevated particle concentration in the presence of TEP.

Of course, if TEP are considered for the calculation of collision rates, they should also be considered for the calculation of particle decrease, e.g. the slope Δ ln C(t)/Δt. This was not done during this experiment, since TEP–CCP aggregates can be either destroyed or artificially generated during the measurement of TEP, severely biasing the results. Recalculated values of α′ might therefore be underestimated.

However, the problem is obvious. In order to differentiate between the probability of particle adhesion and of particle collision due to TEP, TEP–CCP aggregates have to be determined. Alternatively, the parameter α can be considered as a value of the combined probabilities. This is more convenient and appropriate as long as the methodological problems of measuring TEP–CCP aggregates are not solved. Consequently, if this approach is used, the parameter α is no longer limited to values of ≤1.

Particle stickiness during the decline of a diatom bloom

This study showed that coagulation efficiency can be highly variable and may indeed increase during the decline of a diatom bloom. Previously, Kiørboe et al. (Kiørboe et al., 1994) and Dam and Drapeau (Dam and Drapeau, 1995) observed that the stickiness of particles decreased during the course of diatom blooms. One reason might be that their studies took place at the beginning and at the peak of a bloom rather than during the senescent phase.

The increase in apparent particle stickiness was significantly related to an increase in the concentration of TEP relative to CCP. An increase in TEP concentration during in situ phytoplankton blooms has also been observed (Passow and Alldredge, 1994; Mari and Kiørboe, 1996). Thus, the formation of aggregates at the end of diatom blooms may in fact be due to enhanced coagulation efficiency in the presence of TEP. This result corroborates the finding of Logan et al. (Logan et al., 1995) that aggregate formation at the end of phytoplankton blooms coincides with short half-lives of TEP.

Absolute values of α attained during this study were in the upper range of previously reported data. It has been shown that bubbling, such as in the culture vessels used here, may artificially generate TEP (Mopper et al., 1995; Zhou et al., 1998). However, volume concentrations of TEP as attained here were well within the range of in situ values determined in the Baltic Sea by Mari and Burd (Mari and Burd, 1998). Using the Couette Chamber approach, Kiørboe and Hansen showed that coagulation efficiency was linearly dependent on the shear rate (Kiørboe and Hansen, 1993). The shear rate applied here corresponds to values estimated for moderate mixing within the upper ocean layer (MacKenzie and Legget, 1993) and was, with G = 0.86 s^–1, lower than in previous studies. Kiørboe et al. applied a shear rate of G = 30 s^–1 in an investigation of phytoplankton coagulation during a spring bloom and yielded values for α in the range of 0.01–0.136 (Kiørboe et al., 1994). Dam and Drapeau used G = 5 s^–1 during a mesocosm study of the aggregation of a diatom bloom and obtained values of α between 0 and 0.8, more comparable with the values attained here (Dam and Drapeau, 1995). They ran one experiment with G = 10 s^–1, but found no difference in α. However, depending on its intensity, shear can either increase particle collision or increase particle destruction. Riebesell observed maximal aggregate formation after the decline of a diatom spring bloom, while mean shear rate in the upper mixed layer was 0.37 s^–1 (Riebesell, 1991a). The coagulation of exopolymer carbohydrates such as TEP is mediated via cation bridging between the glycoside backbones of the exopolymers (Leppard, 1995). Ramus and Kenney indicated that high shear (>10 s^–1) disrupts microalgal polymer chain interactions, resulting in disaggregation of co-polymers (Ramus and Kenney, 1989). In a similar way, high shear might also disrupt the bonding between free exopolymer particles like TEP and carbohydrates that cover the phytoplankton cell. Thus, coagulation efficiencies under natural low shear might be higher than previously assumed.

Particle stickiness during this study was negatively related to the total abundance of diatoms as well as to single diatom species. In particular, coagulation efficiency was lowest at the highest concentration of CCP at the peak of the bloom. This has also been observed by Kiørboe et al. (Kiørboe et al., 1994) and Dam and Drapeau (Dam and Drapeau, 1995), and confirms the suggestion that particle aggregation is driven by TEP–CCP binding rather than by coagulation of CCP alone. Kiørboe and Hansen hypothesized that S.costatum might escape coagulation by the release of special substances (Kiørboe and Hansen, 1993). Although S.costatum did stick to TEP during the stationary phase of this experiment, there is the probability that healthy cells exudate substances that counteract the formation of TEP.

Conclusions

The coagulation efficiency of particles can be highly variable for a short time during the decline of a diatom bloom. TEP can control coagulation efficiencies significantly and are most likely largely responsible for the formation of aggregates as the bloom ends.

If TEP are present, the apparent particle stickiness (α) calculated from particle abundance using electronic counting systems combines the probability of particle adhesion and of particle collision due to TEP, and can exceed a value of 1. In this case, the probability of aggregate formation might be high even under low concentrations of CCP.

This may have major implications for oligotrophic regions, where the standing stock of phytoplankton is low. Thus, for the understanding of aggregate formation at different oceanic sites, all particles that are involved in coagulation processes have to be determined.

Since coagulation efficiencies may be influenced by the shear rate, the experimental set-up to measure ambient particle stickiness should correspond to the natural environment of the particles studied.

Table I.

Mean abundance of major diatom species and their correlation with the apparent particle stickiness (α) during the laboratory experiment. α′ is a new stickiness parameter calculated by adding the volume fraction of TEP to the volume fraction of Coulter Counter-detectable particles

Major diatom species	Mean abundance (no. cm–3)	rα	Pα	rα′	Pα′
Skeletonema costatum	3963	–0.57	0.04	–0.61	0.04
Chaetoceros sp. <15 μm	477	–0.71	0.006	–0.78	0.003
Thalassionema nitzschioides	399	0.49	n.s.	0.45	n.s.
Chaetoceros sp. <8 μm	376	–0.84	<0.001	–0.85	<0.001
Leptocylindricus danicus	125	0.06	n.s.	0.16	n.s.
Chaetoceros sp. >15 μm	103	–0.52	0.07	–0.55	0.07
Nitzschia sp.	75	0.02	n.s.	0.04	n.s.
Cerataulina bergonii	36	0.23	n.s.	–0.10	n.s.
Leptocylindricus minimus	32	0.049	n.s.	–0.006	n.s.
Rhizosolenia sp.	14	–0.50	0.08	–0.59	0.04
Σ diatoms	5611	–0.66	0.01	–0.71	0.01

Fig. 1.

Decrease in nitrate (a), Chl a (b) and diatom abundance (c) during the course of the 18 day laboratory experiment. Bars are ± 1 SD.

Fig. 2.

Changes in particulate organic carbon [POC; (a)] and transparent exopolymer particles [TEP; (b)] during the decline of the diatom bloom. Bars are ± 1 SD.

Fig. 3.

Numerical abundance [no. cm^–3; (a) and (b)] and volume concentration [φ; p.p.m.; (c) and (d)] of TEP and Coulter Counter-detectable particles (CCP) during the decline of the phytoplankton bloom. Error bars are ± 1 SD (for CCP, n = 3; for TEP, n = 2).

Fig. 4.

Apparent particle stickiness (α) during the decline of the diatom bloom. Values of the new stickiness parameter α′ were calculated by adding the initial volume concentration of TEP to the initial volume concentration of solid particles.

Fig. 5.

Linear relationship between the apparent particle stickiness (α) and TEP:Chl a ratios, withα = 6.38 × 10^–4 (TEP:Chl a) – 3.3 × 10^–3 (r² = 0.92, n = 10).

Special thanks to Avan Antia and Uta Passow for their advice and good suggestions on the manuscript. Thanks are also due to Klaus von Bröckel, Markus Schartau and two anonymous referees for valuable comments on the manuscript. Uschi Junghans and Peter Fritsche are gratefully acknowledged for technical assistance.