INVESTIGATING THE INTRASPECIFIC EFFECT OF CELL CONCENTRATION IN MEDIATING OXYRRHIS MARINA SWIMMING BEHAVIORS

Heterotrophic protists are known to respond to a multitude of abiotic and biotic stimuli which confers a strong selective advantage in marine environments that are frequently dilute and heterogeneously distributed. In this laboratory study, we investigated the role of intraspecific signals in mediating Oxyrrhis marina swimming behavior that could be utilized to enhance dispersive behaviors and reduce competition between intraspecific predators. Using video and image analysis, three-dimensional movement behaviors of O. marina (on scales of micrometers and seconds) were simultaneously quantified with population-scale vertical distributions (on scales of centimeters and hours) and used in dispersal and encounter rate estimates. Three different concentrations of O. marina were filmed in both the absence and presence of the prey alga species, Isochrysis galbana, in at least triplicate films every 30 minutes for three hours at five horizons in 1-L experimental tanks. We found that the cell-cell interactions in the absence of prey cells resulted in modified swim behaviors that increased model estimates of encounter rates by 9%; however, individual swim behaviors between treatments were not significantly different in the presence of prey cells. Also, the relative proportion of the population near the top of the tank significantly decreased by 22% and 16% in both the absence and presence of prey cells, respectively, from low to high O. marina concentrations. These results suggest that O. marina can respond to the intraspecific cell concentration in the absence of competing signals which can ultimately result in significant changes to distributions, growth and grazing rates.


INTRODUCTION
Heterotrophic protists are single-celled microzooplankton that are ubiquitous in the global ocean and are highly diverse in terms of size, taxonomy and feeding behavior (Sherr & Sherr, 1994). They play a vital role as consumers of primary production and dominate trophic interactions at the base of marine food webs, accounting for 60-70% of daily phytoplankton consumption (Calbet & Landry, 2004). By collecting minute prey into larger consumable particles and serving as prey items for larger macrozooplankton, such as copepods, microzooplankton contribute to the availability of food to higher trophic level organisms which ultimately affect the rate of export production (Calbet & Saiz, 2005;Caron & Hutchins, 2013). Therefore, it is important to identify the factors controlling protistan growth and grazing which are key carbon cycle processes that influence primary production, atmospheric carbon exchange and carbon sequestration of dissolved organic carbon to the deep ocean and sediments (Seymour, et al., 2009;Davidson, et al., 2011). There have been many studies that have identified a number of biotic and abiotic factors such as prey cells (Menden-Deuer & Grünbaum, 2006;Martel, 2006), light (Jakobsen & Strom, 2004;Hartz, et al., 2011), and nutrients (Breckels, et al., 2010) that influence population level growth and grazing rates, but there are very few quantitative studies that have focused on the influential factors mediating microzooplankton swimming behaviors. We are just beginning to understand the pace at which microorganisms can respond to changes in environmental conditions and the associated impacts to the food web structure (Kim, et al., 2011;Caron, et al., 2012).
On the microscale level, grazing doesn't result from passive physical encounters between predator and prey, as these microorganisms operate in low Reynolds number environments where viscous forces are dominant and molecular diffusion of particles is significant (KiØrboe, 2008). In addition to a diffusion dominated environment, there is substantial evidence for pervasive heterogeneity on all scales, including the microscale at which plankton operate, which has substantial implications for the rates of encounter between predator and prey (Haury, et al., 1978;Fenchel, 2002;Menden-Deuer, 2008;Durham & Stocker, 2012). Furthermore, most marine environments are extremely dilute and plankton will typically account for a very small percentage of suspended constituents, usually less than 10 ppm by volume (Wolfe, 2000). Therefore, a major challenge for heterotrophic protists, as well as for phytoplankton, is to efficiently locate resources at sufficient concentrations to survive (Caron, et al., 2012). Motile heterotrophic protists have adapted a wide range of behavioral responses which are utilized under different environmental circumstances to maximize foraging efficiency, such as in the absence or presence of prey patches (Montagnes, et al., 2008). One such strategy involves the interpretation of the sharp physicochemical gradients associated with prey patches to direct and modulate predatory swimming behaviors which can increase rates of encounter between predator and prey. The enhanced capacity for heterotrophic grazers to actively search out and exploit these plankton rich patches, can result in a heterogeneous distribution of predator biomass on the order of minutes (Fenchel & Blackburn, 1999), subsequently challenging models that assume constant predatory consumption rates on small (minutes) temporal scales. Therefore, in order to gain a more complete and mechanistic understanding of the planktonic predator-prey interaction and improve modeling efforts, quantitative investigations into environmental signals, such as the role of intraspecific cell concentration, are needed to adequately conceptualize this major trophic pathway in the marine food web (Sherr & Sherr, 2007).
While the quantitative study of signaling is still in its infancy, it has been established that all organisms, whether dead or living, release chemicals into their surrounding environment which are potentially available to be interpreted by any organism with the correct machinery to receive and process such information (Vos, et al., 2006). There have been many laboratory studies that have observed the quantitative changes in both predator and prey swimming behaviors in response to infochemicals. In Menden-Deuer and Grünbaum 2006, Oxyrrhis marina responded to the exuded chemical cues from thin layers of Isochrysis galbana by modulating their pair of constantly beating flagella that decreased vertical velocities and increased turning rates in order to remain in position to exploit this prey-rich area. Behavioral responses to chemical cues have also observed in some motile prey species. When subjected to predatorderived cues, Heterosigma akashiwo increased fleeing behaviors, which resulted in reduced encounter rates and a net positive population growth, as opposed to a net negative population growth when fleeing was not an option (Harvey & Menden-Deuer, 2012). The results of these studies suggest that both prey and predator-derived cues are significant in mediating swimming behavior in autotrophs and heterotrophs.
Oxyrrhis marina was an ideal candidate for our study for a number of reasons. First, it is a highly studied species (reviewed in Lowe, et al., 2011) and its feeding and foraging behaviors have been well-characterized in the literature, which provided context in which to interpret hypothesized modifications of swimming behaviors in response to environmental cues. Second, O. marina is maintainable at high cell densities in culture due to its ability to tolerate a range of conditions and prey sources (Boakes, et al., 2010;Lowe, et al., 2011). Third, the helical swimming trajectories exhibited by O. marina are mainly linear and continuous (Cosson, et al., 1988) which makes it a suitable candidate for establishing a standard 3D framework via video microscopy to quantify swimming behaviors at the individual level. A causal and mechanistic understanding of individual interactions at this level are necessary to establish the basis for population level models that aim to study more representative, and often more complex, scenarios (KiØrboe, 2008). Lastly, O. marina has the potential to serve as a model species to be incorporated into future multi-tropic level behavioral models (Mariani, et al., 2008;Davidson, et al., 2011 (Visser & KiØrboe, 2006). This raises the question, do individuals behave differently in the presence of intraspecific competitors than in the presence of prey? Here, we investigate the role of intraspecific signaling by (1) quantifying the individual swim behaviors of O. marina at three different intraspecific cell concentrations in the absence and presence of a competing prey signal, and (2) associating these individual-level changes with the resulting population distributions and estimates of dispersal and encounter rates.

Culture conditions of predator and prey -The heterotrophic protist,
Oxyrrhis marina (CCMP3375), was cultured in triplicate in 29.6 psu, 0.2 μm sterile-filtered autoclaved seawater (SFSW) collected from Narragansett Bay and incubated at 15°C under low light conditions (~10 µmol photons m -2 s -1 ) on a 12 hour light: 12 hour dark cycle. Cultures were not axenic and fed every 4-5 days with 80 mL of the haptophyte prey alga Isochrysis galbana (CCMP1323) which was grown in SFSW, enriched with f/2 nutrients minus silica (Guillard, 1975 to 29.6 psu) was established using a peristaltic pump in each of the three, 30 cm x 5.5 cm, 1-L octagonal filming tanks to create a stable filming environment by suppressing otherwise dominant water movements associated with convection..
The same source SFSW used to maintain cultures was used to fill the tanks. The filming tanks were covered and held in a temperature controlled room to prevent temperature and air pressure changes from destabilizing the density gradient.
These were essential steps for optimal viewing conditions and the digital reconstruction of the microscale planktonic swimming tracks used to calculate swimming statistics and compare the treatment effects. The methods for video capture were followed and adapted from Menden-Deuer & Grünbaum (2006)  The raw tracks were smoothed by taking 0.1 second subsamples and these smoothed 3D tracks were used to calculate the four aspects of swimming behavior outlined in the following section. Only tracks with a minimum length of 3 seconds were used in the calculation of swimming behaviors.
Statistical analysis -The non-parametric Kruskal-Wallis test was used to determine significant differences (p < 0.05) in swimming behaviors by comparing the mean group ranks of turn rate (degrees second -1 ), vertical velocity (μm second -1 ), swimming speed (μm second -1 ) and vertical deviation angle (degrees) between treatments. Post hoc, one way ANOVA tests were conducted to identify the specific treatments that had significantly differences in individual swimming behaviors (Tukey-Kramer, p < 0.05). All analyses were performed in Matlab, where ν is the effective movement speed (μm s -1 ) and τ is correlation time scale (s). These two parameters were estimated from a least squares regression curve fit of the average root mean square distance versus time. Due to the lack of a significant number of sufficiently long trajectories, our correlative timescale did not extend beyond 30 seconds.
Encounter rates between O. marina and prey cells -To understand the potential implications of the observed shifts in aggregative swim behaviors in response to an enhanced intraspecific signal, we calculated encounter rates as a function of the total volume swept clear by O. marina using the following model from Gerritsen & Strickler (1977).
We used a predator detection radius (R) of 10 μm, which was the sum of the radii for both O. marina cells (~8 μm) and I. galbana (~2 μm), and assumed that predator swimming speeds (v) were much greater than the swimming speed of I. galbana (u), which is known to be a weak swimmer. Therefore, we used a prey swim speed (u) of 5 μm s -1 and our observed predator (v) swimming speeds (μm s -1 ), and a prey concentration of 10,000 cells mL -1 for treatments with added It was determined through one-way ANOVA testing that the horizon depth within each filming tank and differences between replicate tanks were not significant in mediating any of the four analyzed aspects of swimming behavior.  which was significantly slower than the individuals observed in the corresponding no-prey medium and high concentration treatments by 8.4% and 6.8%, respectively (Table 2). In the presence of prey, the turning rates among different O. marina concentrations treatments with prey cells did not significantly differ as median turn rates ranged from 66 to 71 degrees s -1 . For all treatments, the fastest turning rates were most frequently observed within the first 30 minute, which was followed by a sharp decrease in the next 30 minute interval ( Figure 2A, 2B, Table 3). This large temporal variation within treatments diminished within the first hour of observation and the inclusion of these time points did not result in significantly different mean group ranks.

Individual O. marina swimming behaviors -
The mean group ranks of swimming speed (μm s -1 ) were significantly different across all treatments (p << 0.0001) and increased significantly by 8. Therefore, we hypothesize that a strong intraspecific signal could serve as a useful stimulus in the absence of other environmental cues to direct predators towards the surface in an attempt to encounter areas of elevated prey density. however, this behavior may not be completely unexpected. For example, processes such as cell growth, cell proliferation and cell death can be dependent on the local cell concentration and has been demonstrated in a number of multicellular organisms (SØren, et al., 1997). However, aggregative conditions also carry detrimental effects such as increased risk of predation from higher trophic level predators, increased competition for food and increased risk of population wide subjugation to harmful conditions (Schuech & Menden-Deuer, 2014). In terms of this trade-off, our results suggest that O. marina favors the short-term benefit of increased prey encounters within a prey patch over the long-term risks associated with remaining aggregated.

Recognition of prey in the presence of other competitors
Additionally, O. marina swimming behavior has been observed to vary with prey concentration. At high prey concentrations (~10 4 -10 5 cells mL -1 ), longitudinal flagellum (associated with higher swim speeds) have been observed to beat more frequently as compared to low prey concentrations (~10 1 -10 3 cells mL -1 ) where the beating of the transverse flagellum was more frequent which is related to higher turning rates . While the exact predatorto-prey ratio was not calculated throughout the 3 hour observational period, O.
marina consumed I. galbana, and reduced the prey concentration below the 10,000 cells mL -1 threshold suggested by Roberts, et al. (2011), which would predict a simultaneous decrease in swim speeds and increase in turn rates over time. However, our results do not agree with these observations as average swim speeds increased and turn rates decreased. We hypothesize that the transition in individual swim behaviors on the 3 hour time scale were dependent on the absence or presence of prey signals, rather than the actual concentrations of prey, and were facilitated by a shift behaviors as starved O. marina consumed prey.

Aggregative behaviors in the presence of intraspecific signal and prey cells -
We observed an overall decrease in RMSD in nearly all treatments over time, which is characteristic motile behavior for biological organisms which balances increased encounters with prey while mitigating predation risk from higher order predators (Visser, 2007;KiØrboe, 2008). These retentive swimming behaviors are further enhanced by the presence of prey exudates, or the excreted chemical cellular material, as the distance that potential consumers can perceive prey is increased (Larsson & Dodson, 1993). There is increasing evidence to support that O. marina has surface receptors that bind to these prey-derived chemical cues, comparable to the signal transduction pathway observed in the model freshwater protist, Paramecium tetraurelia (Hartz, et al., 2008); however, it is not known if other signals are similarly interpreted (SØren, et al., 1997;Breckels, et al., 2010). Due to time constraints, we did not characterize the intraspecific signal as mechanical, chemical or a combination of both, but O. marina exudates could serve as an effective stimulus to decrease encounter with intraspecific competitors. Theory predicts that in environments with high intraspecific signals that dispersive, ballistic motile behaviors would increase the distance between predators, benefitting the individual by simultaneously decreasing encounters with competitors while increasingly encounters with prey patches. Our results suggest that O. marina did not increase dispersive behaviors in the absence of prey cells at high concentrations of intraspecific cells and were observed to increase retentive behaviors. This is a puzzling and largely counterintuitive response as starved cells in this environment would have been subjected to the greatest competition and presumably would have modified behaviors to increase dispersal rates between competitors. One possible interpretation for this observation is that since it is likely that the intraspecific signal has a chemical component (Vos, et al., 2006) and O. marina is known to have a strong chemotactic response to prey patches (Durham & Stocker, 2012), it is possible that the surface receptors or signal transduction pathways of O.
marina are more generalized which would allow the interpretation of a greater variety of signals at the expense of forming specialized behavioral responses in the presence of multiple signals. This would still permit for well-known prey selectivity through physical encounters between predator and prey (Montagnes, et al., 2008), but does reaffirm doubts concerning O. marina's ability to differentiate chemical signals emanating from mixed assemblages (Martel, 2006).
Further testing is needed to determine the validity of these theories which could be achieved by studying the chemotactic response of O. marina to intraspecific exudates and a deeper investigation into the internal mechanisms used to interpret external chemical cues.

Evidence of unicellular group behavior in protists? -The ability to interpret
intraspecific cues is significant and can serve as the hypothetical basis for coordinated group behaviors, a strategy typically associated with larger multicellular organisms that function to benefit the overall population through the enhancement of specific individual level behaviors. Coordinated behaviors within intraspecific populations has yet to be effectively demonstrated in protists, but has been observed in other microorganisms, most notably in bacteria with regards to quorum sensing (Crespi, 2001). This form of cell-to-cell communication allows bacteria to interpret local conditions (e.g. community composition, strength of chemical cues) and modify individual cell behaviors which has implications at the population-level (Waters & Bassler, 2005). A communicative mechanism that signals the use of a specific set of swim behaviors in O. marina that increase the encounter rate with prey cells would be particularly advantageous during foraging. The topic of protistan group behavior has yet to be thoroughly investigated and O. marina is not known to designate specialized roles within populations, even though each cell presumably have particular swim behaviors that are employed under certain favorable environmental conditions. The few existing studies that have investigated this topic tend to sit at the precipice of what defines group behavior. For instance, Pfiesteria, a single-celled dinoflagellate species was observed to simultaneously release toxins to ambush their prey, which resulted in a large scale fish kill and allowed the dinoflagellates to feed on the carcasses (Burkholder, 1999). The Consequence of aggregative behaviors for encounter rates -The model of Gerritsen and Strickler (1977) provided a useful mechanism to compare encounter rates based on intraspecific variations in swimming behaviors. Over the 3 hour observational period, the approximate 20 μm s -1 increase in median swimming speed across low to high O. marina concentrations in the absence of prey resulted in a 9% increase in the volume swept clear. This simplified model does not account for increased encounter rate due to turning rate (Visser & KiØrboe, 2006) or an enhanced detection radius of predators through interpretation of chemical cues (KiØrboe, 2008). However, the de-correlation length scale (mm) was far greater than the detection radius (μm) between predator and prey, so it is likely that modulations in swimming speeds alone could account for a significant increase in encountered water volumes and prey cells that ultimately influence predation pressure (Harvey, et al., 2013). Oxyrrhis marina have been observed to have maximum ingestion rates of 250 I. galbana cells flagellate -1 day -1 (Goldman, et al., 1989), which would require a prey  (Woodson & McManus, 2007). Predatory foraging behavior may be enhanced by any number of environmental signals, particularly those emanating from prey; however, if swimming behaviors are influenced by the local environmental signals and the behavioral response occurs rapidly, then swim behaviors could be predicted based on specific environments (Visser, 2007). In order to fully describe the range of possible behavioral capacity of O. marina, and possibly many other species of heterotrophic protists, the intraspecific signals between predators should be taken into consideration as an inherent stimulus of swimming behaviors.  Table 2. Summary of medians, percentiles, interquartile ranges and whisker values for box plots reported in Figure 1, for low, medium and high O. marina concentrations in the presence (+) and absence of prey cells (-). The interquartile ranges (IQR) are the difference between the 25th and 75th percentiles. Low whiskers and high whiskers were calculated as Q1 -(1.5 x IQR) and Q3 + (1.5 x IQR), respectively. Data points outside of this range were identified as outliers.  Table 4. Summary of dispersal rates (μm 2 s -1 ), the parameters motility parameters of effective movement speed, ν (μm) and correlative timescale, τ (seconds) derived from the least squares regression curve and the associated goodness of fit (r 2 ) with the average root mean square distance (RMSD).        The distributions of the remaining 150 minutes in the presence of prey varied less significantly than the distributions in the absence of prey, but we still observed a 16 ± 10% lower proportion of the population from low to high O. marina concentrations. Error bars represent one standard error of the mean for triplicate films.