sign in sign up
<<
Home , Bacterivore, Grazing (behaviour), Mixotroph

Ecology, 0(0), 2019, e02874 © 2019 by the Ecological Society of America

Intraguild enables coexistence of competing in a well-mixed water column

1,2,3,4 1 1 HOLLY V. M OELLER, MICHAEL G. NEUBERT, AND MATTHEW D. JOHNSON 1Biology Department, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts 02543 USA 2Biodiversity Research Centre, University of British Columbia, Vancouver, British Columbia V6T 1Z4 Canada

Citation: Moeller, H. V., M. G. Neubert, and M. D. Johnson. 2019. enables coexis- tence of competing phytoplankton in a well-mixed water column. 00(00):e02874. 10.1002/ecy.2874

Abstract. theory predicts that when two compete for a single, finite resource, the better competitor should exclude the other. However, in some cases, weaker competitors can persist through intraguild predation, that is, by their stronger competitor. , species that meet their carbon demand by combining and phago- trophic heterotrophy, may function as intraguild predators when they consume the with which they compete for light. Thus, theory predicts that mixotrophy may allow for coexis- tence of two species on a single limiting resource. We tested this prediction by developing a new mathematical model for a unicellular and phytoplankter that compete for light, and comparing the model’s predictions with a laboratory experimental system. We find that, like other intraguild predators, mixotrophs can persist when an is sufficiently productive (i.e., the supply of the limiting resource, light, is relatively high), or when species interactions are strong (i.e., attack rates and conversion efficiencies are high). Both our mathematical and labora- tory models show that, depending upon the environment and species traits, a variety of equilib- rium outcomes, ranging from competitive exclusion to coexistence, are possible. Key words: ecology; competition; Micromonas commoda; mixotrophy; model-data comparison; Ochromonas.

example, Hutchinson’s description of the “Paradox of INTRODUCTION the ” notes the surprising diversity of phyto- Competition among species for limited resources plays plankton communities despite the fact that their member a fundamental role in structuring ecological communi- species appear to occupy the same resource niche delim- ties. All else equal, when two species are limited by the ited by the availability of light and abiotic resources same available resource, the species that can persist at (Hutchinson 1961). In part, some of this surprisingly the lowest level of the resource when grown in isolation high diversity can be explained by a subtler understand- will competitively displace the weaker competitor ing of resource partitioning among taxa (e.g., use of dif- (Tilman 1977, 1990). This R theory (where R is the ferent forms of nitrogen in phytoplankton; Moore et al. resource-availability threshold required for a species to 2002) and by non-equilibrium dynamics (as was Hutchin- persist) predicts that the composition of an ecological son’s original point; see Hutchinson 1941, 1961). community can be understood by identifying the Trophic interactions can also enhance community resources that limit each member species. The same the- diversity above resource-based expectations. In addition ory holds for light limitation in planktonic communities: to keystone predators, which exert top-down control on “ Huisman and Weissing (1994) derived Iout, the critical community composition by feeding on competitively light intensity” to which phytoplankton superior community members (Paine 1969), intraguild draw down available light in a well-mixed water column. predators, species that eat their competitors, may also In their analysis, the phytoplankter with the lowest Iout increase community diversity (Polis et al. 1989, Polis competitively excludes all other species. and Holt 1992). Specifically, an intraguild predator that However, the observed diversity of some communities is otherwise a weaker competitor may persist in a com- exceeds predictions grounded in resource theory. For munity by consuming its competitors (Holt and Polis 1997), a mechanism that has been empirically demon- Manuscript received 20 August 2018; revised 9 July 2019; strated in several cases (Diehl and Feissel 2001, Borer accepted 15 July 2019. Corresponding Editor: Jef Huisman. et al. 2003, Price and Morin 2004, Wilken et al. 2013). 3 Present address: Department of Ecology, Evolution & , University of California, Santa Barbara, Santa Intraguild predation is widespread in biological systems, Barbara, California 93106 USA. from Diehl and Feissel 2001, Morin 1999) to 4 E-mail: [email protected] insects (Wissinger and McGrady 1993, Borer et al. 2003)

Article e02874; page 1 Article e02874; page 2 HOLLY V. MOELLER ET AL. Ecology, Vol. xx, No. xx to (Fedriani et al. 2000), and most attention (i.e., the availability of sunlight). We then has been focused on taxa that compete for a living prey compared our theoretical results with experiments using item as opposed to a common, abiotic resource (Arim and two widely distributed marine planktonic taxa. Our Marquet 2004). In contrast, in planktonic communities, analysis highlights the generality of intraguild predation mixotrophy is a common form of intraguild predation by as a mechanism for coexistence and enhanced diversity which predators and their prey compete for an abiotic among organisms that share a resource niche. resource (Thingstad et al. 1996, Wilken et al. 2014b). Mix- otrophic species, which combine photosynthesis with pha- The model gotrophic heterotrophy, come in two forms: (1) that ingest prey are termed constitutive mixotrophs because To study the effects of intraguild predation on the they maintain and regulate their own permanently incor- coexistence dynamics of two competing phytoplankton porated plastids. (2) that host symbiotic algae or species, we modified the classic model of Huisman and steal their plastids are known as non-constitutive mixo- Weissing (1994) for phytoplankton living in a well-mixed trophs because, in the absence of prey, they lack photosyn- water column (i.e., each cell experiences, on average, the thetic machinery (Flynn and Hansen 2013, Mitra et al. same amount of light). At a given depth z, the - 2016). Constitutive mixotrophs (referred to as “mixo- specific growth rate gi of a phytoplankter of species i is trophs” hereafter) function as intraguild predators when determined by the difference between its photosynthetic their prey are the phytoplankton with which they compete rate, which is a function of its inherent maximum rate of for light and inorganic nutrients. Mixotrophs are thought carbon uptake via photosynthesis pi and local light avail- to gain a competitive advantage over phytoplankton ability IðzÞ, and its respiration rate ‘i. Thus because they can continue to obtain limiting nutrients, in addition to organic carbon, by (Rothhaupt 1996b, IðzÞ giðzÞ¼pi ‘i (1) Mitra et al. 2016). Their ability to supplement their ener- hi þ IðzÞ getic and carbon needs through photosynthesis also gives them a competitive advantage over pure where hi is the light level at which half the maximum when prey are scarce (Rothhaupt 1996a,Titteletal.2003). photosynthetic rate is achieved. (All variables and To date, most studies of mixotroph persistence have parameters are listed in Table 1). focused on competition for nutrients. A number of theo- Following Huisman and Weissing (1994), we assumed retical studies have modeled mixotroph persistence in that available light declines with depth following the communities that also contain (e.g., phyto- Lambert-Beer law and that each species has a per-cell plankton) and heterotrophs (e.g., ; Crane and light absorptivity ki. Because the water column is well- Grover 2010, Cropp and Norbury 2015, Thingstad et al. mixed, the cell density wi for each species does not 1996), leading to the prediction that mixotrophs should depend on depth. As a result of absorption, the light ð Þ become dominant in in which nutrients are intensity I z decreases with depth from the incident limiting (Mitra et al. 2016). This prediction is consistent level Iin at the surface according to with the observation that mixotrophs are particularly ð Þ¼ ½ð þ Þ : abundant in open ocean oligotrophic gyres (Hartmann I z Iin exp k1w1 k2w2 z (2) et al. 2012) and in coastal ecosystems where a single Integrating over the total depth of the water column, nutrient, such as phosphorus, is scarce (Burkholder the rate of change in each of the two species’ total bio- et al. 2008). However, Rothhaupt (1996a) showed that mass W and W is given by the persistence of mixotrophs alongside strictly hetero- 1 2 trophic competitors depended upon the availability of dW p W light and food as joint-limiting resources. 1 ¼ 1 1 dt k W þ k W Here, we instead focus on competition between mixo- 11 2 2 þ (3) trophs and their intraguild prey for sunlight. In keeping h1 Iin ‘ ln þ ½ðÞþ 1W1 with intraguild predation theory, mixotrophs should h1 Iin exp k1W1 k2W2 gain a competitive advantage when light is the sole limit- ing resource because, by consuming their phytoplankton dW2 ¼ p2W2 dt k W þ k W competitors, they reduce competition for light. In this 11 2 2 (4) paper, we report on our test of this prediction using a h2 þ Iin ln ‘2W2: combination of mathematical and experimental appro- h2 þ Iin exp½ðÞk1W1 þ k2W2 aches. First, we constructed a mathematical model for the interaction of a phytoplankter (the intraguild prey) (See Huisman and Weissing 1994 for the complete and a mixotroph (the intraguild predator) in a well- derivation.) Thus, competition between the two phyto- mixed water column with a single limiting resource plankton species is mediated by competition for light. (light). We evaluated how coexistence and competitive We modified these equations to account for intraguild exclusion depend upon the strength of species interac- predation by allowing species 2 (now denoted M for tions (i.e., intraguild predation) and upon ecosystem “mixotroph”) to predate species 1 (denoted W,in Xxxxx 2019 MIXOTROPHY IN PLANKTONIC COMMUNITIES Article e02874; page 3

TABLE 1. Model symbols and their meanings.

Symbol Description Typical units Simulation values Variable W Phytoplankton (intraguild prey) cells/cm2 M Mixotroph (intraguild predator) cells/cm2 t Time d Parameter 2 1 Iin Surface (incoming) light intensity lmol quanta m s Varied 1 pw Maximum carbon uptake rate, phytoplankter d 1 1 pm Maximum carbon uptake rate, mixotroph d 1, 0.3 2 1 9 7 kw Light absorbance of phytoplankter cm cellW 1 10 2 1 9 7 9 7 km Light absorbance of mixotroph cm cellW 1 10 ,5 10 2 1 hw Half-saturation light intensity for phytoplankter lmol quanta m s 200 2 1 hm Half-saturation light intensity for mixotroph lmol quanta m s 200, 250 1 ‘w Mortality rate of phytoplankter d 0.05 1 ‘m Mortality rate of mixotroph d 0.05, 0.1 a Attack rate of mixotroph on phytoplankter lmol quanta m2 s1 Varied 1 b Conversion rate of phytoplankter to mixotroph cellM cellW Varied

keeping with the Huisman and Weissing (1994) original These light levels are called the “compensation irradi- notation for phytoplankton) with an attack rate a and ances” (sensu Huisman and Weissing 1994). As long as biomass conversion efficiency b: Iin exceeds either Iw or Im, an isolated phytoplankton or mixotroph population, respectively, will grow when dW ¼ pwW small. þ dt kwW kmM Huisman and Weissing (1994) previous analysis of the þ hw Iin pure competition model (Eqs. 3 and 4) showed that, ln ‘wW aWM hw þ Iin exp½ðÞkwW þ kmM except for special parameter combinations that produce (5) functionally identical phytoplankton species, competi- tion for light results in competitive exclusion at equilib- dM ¼ pmM rium. Specifically, each species when grown in dt kwW þ kmM reduces light at the bottom of the water  hm þ Iin column to a fixed, species-specific, value Iout.Intwo- ‘ þ : ln mM baWM species cases, the species with the lowest I outcom- hm þ Iin exp½ðÞkwW þ kmM out petes the other. There is no closed-form expression for (6) — — Iout it must be calculated numerically but it depends Note that, for consistency with the Huisman & Weiss- upon all the model parameters except for light absorp- ing original formulation, the units of the attack rate a tivity. When phytoplankton differ in only one trait, the 2 1 1 outcome is straightforward: the species with the highest are in area per time per mixotroph (cm d cellM ; Table 1). Thus our model assumes that grazing is pro- photosynthetic rate, lowest half-saturation intensity, or portional to the integrated population sizes of the phyto- lowest mortality rate is competitively dominant. plankter and the mixotroph, rather than to their In our model (Eqs. 5 and 6), whenever the mixotroph population densities (i.e., that grazing is independent of M is the superior competitor, it competitively excludes mixed layer depth). However, in planktonic communi- the phytoplankter W as long as surface light Iin is suffi- ’ ties, grazing is likely to depend upon absolute concentra- cient for the mixotroph s persistence. Therefore, we con- tion (i.e., in cells/mL). Thus, our model strictly applies fined our analyses to the more dynamically interesting only when the mixed layer depth does not change with situation in which the phytoplankter W is the superior \ time, as is the case in our subsequent analyses. competitor (i.e., where Iw Im, or, equivalently, where all other parameters are equal and pw [ pm or hw\hm or Analysis ‘w\‘m). In this case, the persistence of the mixotroph M is promoted by its intraguild predation. This scenario is Relative competition for light.—In Eqs. 5 and 6, the min- also likely to be the most biologically relevant because imum surface light levels I and I required for the per- w m mixotrophs, which incur the increased metabolic costs of sistence of the phytoplankter and the mixotroph in maintaining two forms of metabolic machinery (Raven monoculture are 1997), are typically thought to be weaker competitors ‘ ‘ ¼ whw ¼ mhm : for light than phytoplankton (Skovgaard et al. 2000, Iw ‘ and Im ‘ (7) pw w pm m Adolf et al. 2006, Crane and Grover 2010). Article e02874; page 4 HOLLY V. MOELLER ET AL. Ecology, Vol. xx, No. xx

Model equilibria and their stability.—Our model (Eqs. 5 mixotroph, is defined mathematically as the M-axis and 6) has four possible non-negative equilibria: (0,0), at intercept of the M interior nullcline, found by setting which no species are present; ðW ; 0Þ, at which only the W ¼ 0 in Eq. 9. M is determined only by surface light phytoplankter persists; ð0; M Þ, at which only the mixo- Iin and the traits that govern the mixotroph’s photosyn- c c troph persists; and ðW ; M Þ, at which the two species thetic growth, pm, km, hm, and ‘m; it is independent of a coexist. and b because it is the of the mixotroph in The of both species fundamen- the absence of prey. The equilibrium point ð0; MÞ is tally depend upon the availability of light. When surface stable when the M-intercept of the W interior nullcline light Iin is lower than Iw, the no-species ð0; 0Þ equilib- is less than M . Because the latter intercept is controlled rium is stable (because there is insufficient light for phy- by a, we can define a mathematically by evaluating the toplankton growth). With more light, the phytoplankter W nullcline at ð0; MÞ: persists. Further increases in light produce coexistence  p h þ I ‘ of the phytoplankter and the mixotroph, and, ultimately, a ¼ w ln w in w : (10) 2 þ ðÞ result in competitive exclusion of the phytoplankter by kmðÞM hw Iin exp kmM M the mixotroph (Appendix S1: Fig. S1). Transitions between the three positive equilibria also when a [ a (solid vertical line, left panel of Fig. 1), the depend upon the species interaction parameters a and b, equilibrium point ð0; MÞ is asymptotically stable. Thus which determine the attack rate of the mixotroph, and a is the minimum attack rate above which the mixo- the conversion efficiency from phytoplankton to mixo- troph M can, depending upon initial conditions, exclude – troph biomass, respectively (Fig. 1). For Iin [ Im (i.e., the phytoplankter W (regions III V, Fig. 1). surface light is sufficiently high that the mixotroph could Second, the monoculture abundance of the phyto- persist in monoculture), the effects of a and b on the sta- plankter, W , is defined as the W-axis intercept of the bility of equilibria can be understood by examining the W interior nullcline, found by setting M ¼ 0 in Eq. 8. relationship between the zero net growth isoclines (“null- As with M, W depends only on phytoplankter photo- clines”) for each species. These nullclines are determined synthetic traits and surface light. The prey-only equilib- by setting dW=dt and dM=dt equal to zero. There are two rium point ðW ; 0Þ is stable if the W-intercept of the M nullclines for the phytoplankter, defined by the equations interior nullcline (which is also determined by a and b; W ¼ 0 and Fig. 1, middle column) is less than W . Therefore, for any given attack rate a, we can compute a conversion ¼ pw efficiency b ðaÞ at which the nullclines share a W-axis 0 þ kwW kmM intercept. For the M nullcline, this intercept occurs at (8) hw þ Iin ðW ; 0Þ; thus, b must satisfy ln ‘w aM; hw þ Iin exp½ðÞkwW þ kmM  þ ¼ pm hm Iin ‘ þ : 0 ln m b aW and two nullclines for the mixotroph, given by M ¼ 0 kwW hm þ Iin expðÞkwW and (11)

pm [ ð Þ 0 ¼ when b b a (dashed line, left panel of Fig. 1), the þ kwW kmM ðW ; 0Þ equilibrium is unstable. In other words, b ðaÞ is (9) hm þ Iin the minimum conversion efficiency above which M ln ‘m þ baW: hm þ Iin exp½ðÞkwW þ kmM invades and persists regardless of initial conditions (regions II, IV, and V, Fig. 1). These nullclines determine the regions in the phase Finally, because b controls the curvature of the M plane for which each species experiences either positive interior nullcline, we define b^ as the value of b at which or negative population growth. On the nullclines them- the nullclines are just tangent to each other: that is, they selves, the species from whose equation the nullcline is transition from having two positive intersections to hav- derived does not grow, and thus the nullclines’ intersec- ing none. Because the nullclines can only have two posi- tions represent the system’s equilibria. When a ¼ b ¼ 0, tive intersections (region IV, middle column of Fig. 1) the interior (i.e., non-axis) nullclines are parallel (Huis- when a [ a, b^ exists, and is a function of a, only when man and Weissing 1994, fig. 4), leading to competitive a [ a. Biologically, b^ðaÞ (dotted line, left panel of exclusion by the phytoplankter because it has the great- Fig. 1) is the conversion efficiency above which M est competitive ability for light. However, changing val- excludes W regardless of initial conditions (i.e., only the ues of a and b can cause the interior nullclines to ð0; MÞ equilibrium is stable; region V, Fig. 1). intersect up to two times, allowing for multiple stable Because the model involves mixed exponentials, we states (Fig. 1, middle column). solved for a, bðaÞ, and b^ðaÞ numerically for each set of The parameters a and b determine equilibrium stabil- parameters and used numerical simulations to check the ity through their effects on the shape of the nullclines. asymptotic stability of predicted equilibria. We found First, note that M, the monoculture abundance of the that there are five distinct regions in the a-b plane Xxxxx 2019 MIXOTROPHY IN PLANKTONIC COMMUNITIES Article e02874; page 5

Population Parameter space Nullclines dynamics 30 . . W = 0 M = 0 I 5 101

1

b 0 30 II 105 II:II: CoexistenceCoex V: M Wins 1 1 (

105 ) 1 6 0 30 III 1 (x10 M 101 5

Conversion efficiency, Conversion efficiency, 1 IV: MSS ((MM or Coexistence)Coex) 0 W , 30 5

Mixotroph, IV I: W Wins III:III: MSSMSS ((MM oror WW)) 101 M

1 )

Attack rate, a (x10–8) 0 105 30 V 1 (W*,0) stable (0,M*) stable (Wc,Mc) stable 1 c c M* W M W* M* 5 (0, ) or ( , ) (,0) or (0, ) 101 a = a* b = b*(a) b = b^ (a) 0 1 0 20 40 60 80 0 10 20 Phytoplankter, W (x106) Time

FIG. 1. The effect of species interaction parameters (attack rate a and conversion efficiency b) on equilibrium population out- comes. When attack rates are low (a\a), the phytoplankter always persists; for sufficiently high conversion efficiencies (b [ bðaÞ), the phytoplankter and mixotroph coexist. When a [ a, the asymptotic population dynamics depend upon b.Asb increases, model stability transitions from bistability (either the phytoplankter or the mixotroph outcompetes the other, depending upon initial conditions) to a different type of bistability (either the mixotroph outcompetes the phytoplankter, or the two species coexist) to a single equilibrium point (the mixotroph outcompetes the phytoplankter). Roman numerals on the leftmost panel corre- spond to a and b parameter choices whose nullclines are displayed in the middle column. Stable equilibria are marked by colored symbols (blue stars = phytoplankter-only stable, red stars = mixotroph-only stable, purple squares = coexistence stable), with corre- sponding example population trajectories shown in the rightmost column. Positive, unstable equilibria are indicated by gray circles. Two nullclines (W ¼ 0 and M ¼ 0) and the unstable equilibrium at ð0; 0Þ are unmarked in each nullcline plot). Parameters are listed 7 in Table 1, with Iin ¼ 100, pm ¼ 0:3, hm ¼ 200, km ¼ 1 10 , and ‘m ¼ 0:05. delimited by the curves a ¼ a, b ¼ bðaÞ, and b ¼ b^ðaÞ mixotroph persistence should increase; and (2) with (Fig. 1). Two of these regions exhibit bistability. That is, increasing attack rates, mixotroph persistence should there are two asymptotically stable positive equilibrium also increase. Further, our model makes specific predic- points, and the dynamics of the system are therefore tions about the number and stability of equilibria in the dependent upon initial conditions (Fig. 1, regions III system for given sets of parameters. To test our model’s and IV; see alternative population dynamics in the right- predictions, we developed a laboratory experimental sys- most column and velocity diagrams and population tra- tem using mixotrophic and photosynthetic plankton, and jectories in Appendix S1: Fig. S2). manipulated light, attack rates, and initial conditions. The location and extent of these stability regions depends upon model parameters. For example, more sur- Study organisms and culture conditions.—Our laboratory face light (larger Iin) increases the portion of a-b parame- experimental system used two taxa: Micromonas com- ter space over which the mixotroph can persist and moda (strain CCMP 2709; Van Baren et al. 2016, Worden competitively exclude the phytoplankter (Fig. 2). Species et al. 2009), a cosmopolitan eukaryotic phytoplankter traits, including half-saturation constants and mortality that is dominant in both coastal and oligotrophic ocean rates, affect the model’s sensitivity to a and b, but do not regions (Cottrell and Suttle 1991, Not et al. 2004), and qualitatively change the trajectory of the system’s Ochromonas sp. (strains CCMP 1391, 1393, and 2951), a response to resource enrichment (Appendix S1: Fig. S3). mixotrophic chrysophyte (Rothhaupt 1996a). Ochromo- nas-like are important grazers in a variety of marine and freshwater environments. Because both Empirical comparison Ochromonas and Micromonas are known to be bacterivo- Collectively, our analysis of Eqs. 5 and 6, leads to two rous (Sanders and Gast 2011, Wilken et al. 2013, McKie- main predictions: (1) With increasing surface light levels, Krisberg and Sanders 2014), we worked with axenic Article e02874; page 6 HOLLY V. MOELLER ET AL. Ecology, Vol. xx, No. xx cultures to ensure that growth was limited to phototrophy for a period of at least 4 weeks prior to beginning any (both species) and intraguild predation (grazing of Micro- data collection. All cell enumeration was done by sub- monas by Ochromonas). Cultures CCMP 2709 and 1391 sampling each culture for analysis by size and fluores- were ordered from the National Center for Marine Algae cence on a Guava easyCyte flow cytometer (EMD and Microbiota (NCMA, Bigelow Laboratory, East Millipore, Darmstadt, Germany). Boothbay, ME, USA), and CCMP 1393 and 2951 were provided by S. Wilken. Quantifying relative photosynthetic growth rates.—To All stock cultures were maintained in K medium (Kel- verify that our experimental system was appropriate to ler et al. 1987; see Table S1 for list of nutrient contents) our mathematical model, we conducted a series of prelim- made by adding pre-mixed nutrients (ordered from the inary experiments. First, we grew both Micromonas and NCMA) to 0.2 lm filtered Santa Barbara coastal seawa- Ochromonas in isolation and compared their maximum ter. Stock cultures were maintained in 40-mL tissue cul- photosynthetic growth rates. Growth rates were measured ture flask batch cultures at 24°C under a 12 h:12 h by inoculating cells at 2,000 cells/mL (Ochromonas)or light:dark cycle with light illumination from above. We 50,000 cells/mL (Micromonas) in triplicate, and quantify- used mesh screening, in combination with varied shelf ing population sizes for a period of up to 8 d. For each proximity to overhead light sources, to create three light replicate, we calculated growth rate as the maximum slope environments (100, 50, and 20 lmol quanta m2 s1) of a semilog plot of population size vs. time. In some and acclimated cultures to their experimental light levels cases, this calculation required eliminating later (i.e., after

0.1 (a) (W*,0) stable (0,M*) stable I c c in = 15 (W ,M ) stable (0,M*) or (Wc,Mc) stable 0 (W*,0) or (0,M*) stable 0.1 (b) a = a* b b a b = *( ) ^ Iin = 20 b = b(a)

0 0.1 (c)

Iin = 50 Conversion efficiency, Conversion efficiency, 0 0.1 (d)

Iin = 100

0 0 0.5 1 Attack rate, a (x10–8)

FIG. 2. Effect of changing light levels on equilibrium population outcomes. Each panel shows the stability of equilibria for a range of attack rates a and conversion coefficients b. Surface light increases from top to bottom. As the resource becomes more enriched (increasing light), the range of parameter space over which the mixotroph may persist increases. For example, the yellow triangles, which represent a system with a ¼ 2 109 and b ¼ 0:05, indicate that, as surface light increases, the system would transi- tion from a phytoplankter-only equilibrium (Iin ¼ 15, 20) to coexistence (Iin ¼ 50) to mixotroph-only (Iin ¼ 100). Equilibrium stability is indicated as in Fig. 1. Xxxxx 2019 MIXOTROPHY IN PLANKTONIC COMMUNITIES Article e02874; page 7

Day 6) time points because population growth slowed as have grazing-enhanced phototrophy (Lie et al. 2018). populations approached . We defined Indeed, we observed that two of our Ochromonas strains maximum photosynthetic growth rate as the growth rate (CCMP 1393 and CCMP 2951) achieved higher popula- achieved at 100 lmol quanta m2 s1 (the highest light tion sizes when prey were available (Appendix S1: level). We found that the phytoplankter Micromonas had Fig. S7). Because carrying capacity increased with a growth rate approximately three times as high as the increasing light levels for all four taxa used in the study, three mixotrophic Ochromonas strains (Fig. 3a; we inferred that light was a limiting resource in our Appendix S1: Fig. S4). In this two-species system, as in experimental system (Fig. 3c). However, we were sur- our model analysis, the mixotroph is the weaker competi- prised to note that, although increases in Micromonas tor for light. populations with light were statistically significant, the magnitude of these changes was smaller than expected Measuring a gradient of attack rates.—We used different (approximately 5%). We also estimated absorption coef- strains of Ochromonas to manipulate the variable of ficients for each taxon by measuring light transmission attack rate. Because, to our knowledge, grazing of through 1-cm of cultures of known concentration, and Micromonas by Ochromonas has not been previously 7 2 found that kw =1 9 10 cm /cell and km = reported, we first measured attack rates (i.e., grazing 5 9 10 7 cm2/cell. We did not observe significant differ- rates) in a separate experiment. To do this, we incubated ences in absorption coefficients for cells acclimated to Micromonas with and without Ochromonas at all three different light levels, or for fed vs. unfed mixotroph cells, light levels and at three predator:prey ratios (1:10, 1:20, so we pooled data across light levels. We used this infor- and 1:100, with a starting Ochromonas concentration of mation to select an experimental water column depth of 2,000 cells/mL and a starting volume of 2 mL), and 8 cm for our experimental test of model predictions. measured changes in population size over 24 h. Over We also tested for the role of other, potentially co-lim- this time period, prey were never completely extirpated, iting resources in our experimental system. We verified and the maximum decrease in prey abundance was by that inorganic carbon was unlikely to be limiting by 50% from starting densities. Thus, it is unlikely that prey measuring pH of fresh media and of 30-d-old stock depletion caused an underestimation of Ochromonas (non-experimental) cultures, and finding no significant attack rates. We used the equations of Frost (1972) and difference (pH fresh: 8.41; pH old: 8.51 0.05; Heinbokel (1978) (developed inJeong and Latz 1994) to P = 0.129, t-statistic = 1.811, df = 5). To prevent nutri- calculate grazing rates for each of the three strains of ents from becoming co-limiting during experiments, we Ochromonas. We found that Ochromonas did graze on ran our experimental test of model predictions in nutri- Micromonas, with rates that differed consistently across ent-rich 2K media (i.e., media with twice the nutrient strains and increased with increasing prey:predator content of K media; Keller et al. 1987). We used pub- ratios (Fig. 3b). We confirmed that grazing rates lished estimates of cellular nitrogen content to estimate increased linearly with prey:predator ratio by fitting the N budget in our cultures (Appendix S1: Table S2) both linear (Holling Type I) and saturating (Holling and found that at maximum population densities, Micro- Type II) models to the data, and comparing the associ- monas and Ochromonas would use approximately 4–20% ated values. AIC values were the same (CCMP 1391 and of the available N in the 2K culture medium respectively 1393) or lower (CCMP 2951) for the linear models, indi- (Appendix S1: Table S3). We also performed a separate cating that a linear (Type I) approximation was a good experiment in which we tested for the effects of nutrients fit for our empirical system. Grazing was confirmed by on carrying capacity by growing Micromonas at 50 using fluorescence microscopy to visualize Micromonas lmol quanta m2 s1 in media with half (K/2), full prey cells inside of Ochromonas digestive vacuoles, and (K), double (2K), or quadruple (4K) the nutrient con- by verifying that Ochromonas growth rates increased tent of K media, and found that carrying capacity was with increasing prey availability in later experiments not significantly affected by nutrient content (Appendix (Appendix S1: Fig. S5). These data allowed us to manip- S1: Fig. S6). ulate the attack rate (i.e., a) by using three strains of Ochromonas with different grazing rates. Generation of comparable model and experimental Testing for light limitation.—We verified that our system data.—We used data on species traits to determine the was light-limited by demonstrating that each organism’s appropriate qualitative comparisons between our mathe- carrying capacity increased with increasing light avail- matical and empirical systems. Specifically, we modeled ability. As part of our main experiment (details below) the expected dynamics for varied surface light Iin and we measured carrying capacity as the maximum popula- attack rates a given other parameters chosen based on tion size achieved by each taxon at each light level empirical data: pm ¼ 0:3 (i.e., the photosynthetic growth over a period of 20 (100 lmol quanta m2 s1), or rate of the mixotroph was 30% that of the phytoplank- 2 1 40 d (20 lmol quanta m s ). For Ochromonas,we ter), hm ¼ 250 (i.e., the half-saturation irradiance of the included data in which Micromonas prey were initially mixotroph was larger than that of the phytoplankter), 7 available but were ultimately eliminated by grazing, km ¼ 5 10 (i.e., the absorptivity of the mixotroph because some Ochromonas strains have been shown to was five times that of the phytoplankter), and ‘m ¼ 0:1 Article e02874; page 8 HOLLY V. MOELLER ET AL. Ecology, Vol. xx, No. xx

(a) (b) A 100

1.2 per CCMP 2951 80 CCMP 1391 1.0 CCMP 1393

0.8 per day) 60

0.6 C Micromonas D 40 0.4 a 20 A

Ochromonas b 0.2 B growth rate (per day) B B b Maximum photosynthetic 0.0 Grazing rate ( 0 Micromonas Ochromonas 1 to 10 1 to 20 1 to 100 CCMP 2709 CCMP 2951 CCMP 1391 CCMP 1393 Ochromonas to Micromonas ratio

Species

(c) 5 2.0 B C A LightLigh level 20 20 µmol·quanta·m50-2·s-1 4 C 100 1.5 50 µmol·quanta·m-2·s-1 B 100 µmol·quanta·m-2·s-1 3

1.0

2

0.5 c 1 b A a Carrying capacity (cells/mL, in millions) 0 0.0 20 50 100 CCMP 2951 CCMP 1391 CCMP 1393 Micromonas Ochromonas

FIG. 3. Traits of the four experimental organisms. Panel A: The phytoplankter Micromonas CCMP 2709 exhibited a signifi- cantly greater maximum photosynthetic growth rate than the three mixotrophic Ochromonas strains CCMP 1391, 2951, and 1393 (P < 0.01, Tukey’s [honestly significant difference]); the three Ochromonas strains also differed, thought not as substantially, from one another (P < 0.05, Tukey’s HSD). Data are from pilot experiments. Panel B: The three Ochromonas strains differed in their grazing rates, with strain 2951 exhibiting the highest grazing rates, and strain 1393 the lowest. Note that empirically measured graz- ing rates are equivalent to aM (the product of the mathematical model’s attack rate and the population of Micromonas). To convert between grazing rates and attack rates used in the mathematical model, one must divide by the population size of Micromonas. Grazing rates increased with increasing initial prey concentrations. Letters indicate significant differences at the P < 0.05 level (Tukey’s HSD comparing grazing rates grouped by prey concentration). Data are from pilot experiments. Panel C: Carrying capaci- ties of the four experimental organisms as a function of light level. Both Micromonas (left) and the three Ochromonas strains (right) exhibited increasing maximum population sizes with increasing input light (Tukey’s HSD comparing population size grouped by species; letters indicate significant differences at the P < 0.05 level). Data are from the main experiment.

(i.e., the intrinsic mortality rate of the mixotroph was bistability of single-species equilibria (Fig. 4). Further- twice that of the phytoplankter). Thus, unlike in our more, in regions of bistability, increasing surface light strictly theoretical exploration (Figs. 1, 2, Appendix S1: reduces the basin of attraction for the coexistence equi- Fig. S3) in which we varied traits independently, our librium (Fig. 4, compare panels II and III), meaning empirical data suggested that Ochromonas strains were that a wider range of initial conditions should lead to competitively inferior to Micromonas because of simul- exclusion of the phytoplankter by the mixotroph. taneous differences in multiple traits. We varied light We tested this prediction by generating time series intensity and attack rate over parameter ranges that cap- population data on Ochromonas and Micromonas in co- tured the most dynamically interesting regions of param- culture. We manipulated the availability of light, the lim- eter space; in both cases, this meant varying these iting resource, using mesh screens as described above, parameters over ranges that were greater than in our and manipulated attack rate by using three strains of empirical system. Ochromonas. Because our mathematical model predicted For a given attack rate, our model predicted transi- bistability for some parameter combinations, we also tions from phytoplankton-dominated to mixotroph- varied experimental initial conditions by initiating exper- dominated equilibria with increasing surface light iments with 1:10, 1:20, or 1:100 ratios of Ochromonas to (Fig. 4). The sequence of transitions depends upon Micromonas. For example, when the initial density of attack rate: For relatively low values of a, increasing Ochromonas was 2,000 cells/mL, we initialized Micromo- light should cause systems to transition from phyto- nas at 20,000, 40,000, or 200,000 cells/mL. To verify that plankter-only, to bistability of single-species equilibria, experimental conditions were viable for population to bistability of coexistence and mixotroph-only, to mix- growth of Micromonas in the absence of competition, we otroph-only states. For higher values of a, there is no also set up three Ochromonas-free controls with initial Xxxxx 2019 MIXOTROPHY IN PLANKTONIC COMMUNITIES Article e02874; page 9 abundances of 20,000, 40,000, or 200,000 Micromonas population densities at the final time point and the 8 cm cells/mL. Note that some variation in initial abundances water column depth. Although Micromonas control and did occur, but ratios were held constant. Micromonas-only populations consistently drew Iout For each parameter and initial condition combination down to low levels, there were no significant differences (3 light levels 9 3attackrates9 3 initial conditions = 27 among Iout levels (Appendix S1: Fig. S9), in contrast to treatments), we grew Ochromonas and Micromonas a predicted decrease in Iout with increasing Iin (Huisman together in co-culture and collected population size data and Weissing 1994), empirically demonstrated by Huis- at intervals of one (100 lmol quanta m2 s1), two (50 man 1999). This was due to the relatively small popula- lmol quanta m2 s1), or three (20 lmol quanta tion response of Micromonas to increasing light levels m2 s1) days until the populations reached equilibrium (Fig. 3c), which may indicate that, despite our other sizes (20, 30, and 40 d, respectively). Each treatment and measurements, Micromonas experienced limitation by a control was run in triplicate. Each experimental replicate non-light resource in some of the experiments. Overall, contained a 10-mL semi-continuous batch culture in a 14- trends for Ochromonas-only and coexistence equilibria mL culture tube. Although the sides of the tubes were followed model predictions, with increasing Iout levels clear, cultures were illuminated from above to improve for coexistence equilibria as attack rates increased model-experiment agreement. Batch cultures were initial- (Appendix S1: Fig. S9). However, Ochromonas popula- ized by inoculating sterile media with a known concentra- tion sizes were sometimes low (Appendix S1: Fig. S7), tion of Micromonas and/or Ochromonas. At each sampling which contributed to divergence from model predictions point, cultures were mixed thoroughly by inversion, and a (especially for prey-free treatments). 250 lL sample was removed for flow cytometry enumera- tion. Culture volume was then replenished by adding DISCUSSION 275 lL of fresh 2K media (the extra 25 lL compensated for low levels of evaporation); this corresponded to low Using a mathematical model for mixotrophy grounded dilution rates of 0.0275 (at high light) to 0.009 (at low in resource competition theory, we demonstrated that light) per day. two plankton species may coexist even when competing for a single limiting resource (light). Consistent with Experimental data are consistent with model predic- other forms of intraguild predation (Holt and Polis tions.—Our population time series data qualitatively sup- 1997, Mylius et al. 2001), mixotrophy permits the persis- port model predictions. At the lowest attack rate (CCMP tence of a weaker competitor for light that would other- 1393, leftmost column of Fig. 5), as surface light wise be competitively excluded (Huisman and Weissing increased, observed population dynamics transitioned 1994). Our results complement the mathematical model- from a Micromonas-only equilibrium to bistability ing work of Thingstad et al. (1996), who also showed between Micromonas-only and Ochromonas-only equilib- that mixotrophs can coexist alongside their prey when ria. At higher attack rates (CCMP 1391 and 2951, center competing for a limiting resource (in their case, nutri- and rightmost columns of Fig. 5, respectively), increas- ents). However, coexistence is not guaranteed: If the ing light drove a transition from coexistence, to bistability mixotroph has a high attack rate or conversion effi- between coexistence and an Ochromonas-only equilib- ciency, or if surface light intensity is high, the mixotroph rium (Fig. 5; see also Appendix S1: Figs. S7 and S8 for can exclude the . This is in keeping with individual replicate time series data). Furthermore, com- other theoretical results that predict mixotroph domi- parisons of initial condition-dependence between inter- nance in high-resource environments (Wilken et al. mediate and high light levels suggest that, consistent with 2014a, Ptacnik et al. 2016) such as eutrophic freshwater, model predictions, the basins of attraction for Ochromo- estuarine, and coastal marine ecosystems. nas-only equilibria are growing with increasing light Our experimental study qualitatively supported our (Fig. 5, compare top and middle rows). mathematical model’s predictions, with the mixotroph However, some interesting contrasts between our Ochromonas becoming more dominant (i.e., more likely empirical and mathematical models emerged. First, our to exclude the phytoplankter Micromonas) with increas- empirical results indicate that at low light levels and ing surface light intensity and increasing attack rates. intermediate attack rates, the Micromonas-only and The increasing of the mixotroph as ecosys- coexistence equilibria were simultaneously stable (Fig. 5, tem productivity increased is consistent with a number middle of bottom row). This contradicted our mathe- of empirical studies of intraguild predator abundance in matical model, which did not predict the simultaneous protistan (Morin 1999, Diehl and Feissel 2001, Price and stability of this pair of equilibria. Second, our empiri- Morin 2004) and insect (Borer et al. 2003) systems. We cally observed Iout light levels at the base of the experi- observed all three types of non-negative model equilibria mental water columns were more variable than the (phytoplankter-only, mixotroph-only, and coexistence) model’s predictions (Appendix S1: Fig. S9). Because our and five types of model dynamics (monostability of each experimental water columns were not amenable to direct of the equilibria, simultaneous stability of the phyto- measurements using our light meter, we instead calcu- plankter-only and mixotroph-only equilibria, and simul- lated Iout by multiplying taxon-specific absorptivities by taneous stability of the coexistence and mixotroph-only Article e02874; page 10 HOLLY V. MOELLER ET AL. Ecology, Vol. xx, No. xx

Parameter space Isoclines VV V 4

2 IVaIVa 0 4 IVa

in IVb I IVb 2 ) 6 III 0 III 4

(x10 IVb M 2

Surface light, II 0 III

Mixotroph, 3

2

0 I

2 Attack rate, a (x10–7) W* M* Wc Mc 0 ( ,0) stable (0, ) stable ( , ) stable 0 20 40 80 (0,M*) or (Wc,Mc) stable (W*,0) or (0,M*) stable Phytoplankter, W (x106)

FIG. 4. Dependence of mathematical model predicted dynamics on surface light Iin and attack rate a. Legends and notation as in Fig. 1. As available light increases, the system transitions from competitive exclusion by prey, to alternate competitive exclusion states or coexistence, to bistability of coexistence and competitive exclusion by the mixotroph, and finally to competitive exclusion 7 by the mixotroph. Nullclines are shown for increasing values of Iin at a fixed value of a = 1 9 10 (right column); Roman numer- als are used to indicate specific parameter values and are chosen for consistency with Roman numerals in Fig. 1. Note that the basins of attraction for the bistable equilibria shift. For example, as Iin increases, the basin of attraction for the coexistence equilib- rium (purple square) shrinks (nullclines corresponding to locations IVa and IVb). Parameters are as listed in Table 1, with pm = 0.3, 7 hm = 250, ‘m = 0.1, km = 5 9 10 , and b = 0.005. equilibria). Surprisingly, we also observed simultaneous competitive inferiority. We expect that expanding our stability of the phytoplankter-only and coexistence equi- analysis to allow tradeoffs among species’ traits would libria (Fig. 5, middle column of bottom row), which was also reveal a complex dependence of competitive superi- not predicted by our mathematical model. We postulate ority on productivity, consistent with Huisman and that this outcome (in which Micromonas competitively Weissing (1994). For example, coexistence may also be excluded Ochromonas when it was inoculated at the mediated by divergent use of light \citep[e.g., by main- highest population densities) may have resulted from taining photosynthetic pigments with absorption spectra stochastic extinction of a small population of Ochromo- that are maximized at different wavelengths; Stomp nas as the populations dynamically approached equilib- et al. 2004) and nutrient resources (e.g., by contrasting rium. nutrient use efficiencies, but seePassarge et al. 2006). The Huisman and Weissing (1994) model for compet- Furthermore, in the Huisman and Weissing (1994) anal- ing phytoplankton also allows for competitive outcomes ysis, outcomes could be predicted on the basis of Iout to depend upon the productivity of an ecosystem. In (which is inversely proportional to Iin) measured on each their model, when phytoplankton differ in multiple traits competitor in monoculture. In contrast, in our study, the (e.g., one species has a larger photosynthetic rate and a mixotroph may competitively exclude the phytoplankter higher half-saturation light intensity than the other), the even when, in monoculture, its Iout is greater, because it competitively dominant species may change as incoming is also able to feed on the phytoplankter (Appendix S1: light levels increase. In our analysis, we focused only Fig. S9). Because competition for light is mediated by on single-trait differences that result in productivity- standing stock biomass rather than production of new independent competitive dominance for the phytoplank- biomass, the competitive interactions between mixo- ter. This allowed us to highlight the role that intraguild trophs and phytoplankton mimic interference competi- predation plays in permitting persistence in spite of tion. Curvature of the corresponding nullclines (e.g., Xxxxx 2019 MIXOTROPHY IN PLANKTONIC COMMUNITIES Article e02874; page 11

Increasing attack rate, a CCMP 1393 CCMP 1391 CCMP 2951

107

107 106

6 10 105 Ochromonas

in 4 I 10 105 X X X X X X X X 022022022022022022022022022

107 abundance (cells) 107 106

6 10 105 abundance (cells)

104 105 X X X X X X 030030030030030030030030030

Micromonas 7 Increasing surface light, 10

107 106

6 10 105

104 105 X X X X 040040040040040040040040040 Experimental day 10:1 20:1 100:1 10:1 20:1 100:1 10:1 20:1 100:1 Prey : predator ratio Prey : predator ratio Prey : predator ratio

FIG. 5. Empirical test of model predictions. We experimentally manipulated light (by conducting experiments at three different light levels, top row = 100, middle row = 50, and bottom row = 20 lmol quanta m2 s1), attack rate (by using three different Ochromonas strains, left column = CCMP 1393 with the lowest attack rate, right column = CCMP 2951 with the highest attack rate), and initial conditions (subplots grouped in threes, with initial conditions of 1:10, 1:20, and 1:100 Ochromonas}:Micromonas from left to right). Time series of mean population sizes with standard errors are plotted for the mixotroph Ochromonas (solid lines) and the phytoplankter prey Micromonas (dashed lines). Y-axes (log scale) are the same for all plots. Extinctions are marked with an “X,” and subplot background color indicates the type of equilibrium observed (blue = phytoplankter only, red = mixotroph only, and purple = coexistence). Our data show a transition from Micromonas-dominated equilibria (phytoplankter-only and simultane- ous stability of alternate competitive exclusion) to Ochromonas-dominated equilibria (mixotroph-only and simultaneous stability of mixotroph-only and coexistence) as grazing rates and light levels increase (from lower left to upper right). This is qualitatively con- sistent with the mathematical model predictions presented in Fig. 4.

Fig. 1) therefore allows for coexistence under specific Our goal in this study was to develop and test a model ranges of parameter values. with a very general representation of intraguild preda- The outcomes of our laboratory experiment also dif- tion among competitors competing for light as a single, fered from model predictions in ways that illustrated the shared limiting resource. Therefore, in our mathematical importance of grazing to mixotroph growth. For exam- model, we disregarded the possibility of other limiting ple, we found that Ochromonas strains CCMP 1393 and resources, such as nutrients, in our system. Although we 2951 often attained higher population sizes when grown did not measure nutrient levels over the course of our with Micromonas prey than when grown in isolation, study, several lines of evidence indicate that light was the even if they subsequently competitively excluded Micro- major limiting factor in our system. First, we initi- monas (Fig. 5 and Appendix S1: Fig. S7). This had ated the experimental cultures with high concentra- implications for light transmission (Iout), as light trans- tions of nutrients in the media (Appendix S1: Table S1; mission for mixotroph-only equilibria tended to be Keller et al. 1987) relative to estimated cellular quotas higher than model predictions (Appendix S1: Fig. S9). (Appendix S1: Table S3). Second, carrying capacity Although all three strains used in this study were capable increased with increasing light availability (Fig. 3c) but of transiently sustaining photosynthetic growth in the was insensitive to media type (Appendix S1: Fig. S6). absence of prey (Fig. 3), other studies have shown that Finally, existing photophysiological studies of M. com- prey can enhance mixotroph photosynthetic capacity, moda (Thompson et al. 1991) and Ochromonas (Wilken perhaps by providing nutrients inaccessible in the culture et al. 2013) indicate that growth rates are light limited media (Sanders et al. 2001, Lie et al. 2018). at irradiances below 100 lmol quanta m2 s1 in both Article e02874; page 12 HOLLY V. MOELLER ET AL. Ecology, Vol. xx, No. xx taxa. The qualitative agreement between our mathemati- cytometer) for Ochromonas regardless of light and prey cal and empirical systems suggests that, indeed, the sin- environment. One possible explanation is that undi- gle-resource model captured a core phenomenon about gested Micromonas chlorophyll inside of Ochromonas the experimental system. However, transcriptomic data digestive vacuoles may have increased per-mixotroph suggest that Ochromonas strain CCMP 1393 may not be pigment estimates, countering reductions in chlorophyll capable of using nitrate (Lie et al. 2018), which was the production by Ochromonas. Additional experiments most abundant source of inorganic nitrogen in our cul- would be needed to test this hypothesis. ture medium (Appendix S1: Table S1). If strains prefer- In the course of our experiments, we obtained, to our entially utilized ammonium, this form of nitrogen could knowledge, the first measurements of marine Ochromo- have been completely depleted (Appendix S1: Table S3), nas grazing on a eukaryotic phytoplankter, though such potentially causing the mixotrophs to function more like observations have been made on freshwater strains (Bor- strict predators and promoting coexistence (Wilken aas et al. 1992). Both Micromonas commoda and Ochro- et al. 2014b). monas are important in planktonic food webs. Incorporating co-limitation by nutrients (Stickney Ochromonas has long been recognized as a et al. 1999, Crane and Grover 2010), as well as other that can exert top-down control on cyanobacteria more complex mechanisms of species interactions such (Wilken et al. 2014a) and heterotrophic bacteria (Kate- as alternate predation functional responses (Holling chakis and Stibor 2006), allowing it to outcompete strict 1959), and light-dependence of photosynthetic (Flynn heterotrophs (Rothhaupt 1996a, Katechakis and Stibor and Hansen 2013) and grazing rates (Skovgaard 1996, 2006). Micromonas is a cosmopolitan eukaryotic pico- Holen 1999, Strom 2001), could improve the predictive phytoplankter that may be dominant in both coastal abilities of a model for a specific system. For example, and open-ocean settings (Cottrell and Suttle 1991, Not although in our study grazing rates did not systemati- et al. 2004). The two species have been observed to co- cally increase with increasing light (sensu Strom 2001; occur at high abundances Furuya and Marumo 1983), see Appendix S1: Fig. S10, especially higher prey ratios), though no studies of potential grazing interactions have correlations between attack rates and surface light yet been conducted. Because the relative contribution of would accelerate the transition to mixotroph-dominated heterotrophy to Ochromonas metabolism is expected to equilibria. Our approach also does not account for increase with warming surface ocean temperatures changes in physiology under resource-limited conditions, (Wilken et al. 2013), understanding the breadth of this such as photoacclimation (Herzig and Dubinsky 1992) mixotroph’s potential prey will be critical to predicting or shifts in reliance on different forms of metabolism changes to planktonic production and nutrient cycling. under different light regimes (Mitra et al. 2016). A number of theoretical studies have considered the Our Ochromonas strains did not exhibit a clear trade- role of constitutive mixotrophy, more generally, in con- off between investing in phototrophic vs. heterotrophic straining the dynamics and function of planktonic com- metabolisms that has been hypothesized to constrain munities. By providing intermediate linkages within and mixotroph ecology: Although strain CCMP 2951 had between food chains, mixotrophs may serve as stabilizers the highest grazing rate and lowest photosynthetic rate, of community dynamics (Jost et al. 2004, Hammer and CCMP 1391 had the second highest grazing rate, yet Pitchford 2005) and as important mediators of the trans- highest photosynthetic rate (Fig. 3). Further, strain fer of to higher trophic levels (Ptac- CCMP 2951 achieved the highest mixotrophic growth nik et al. 2004) and, potentially, carbon export (Ward rates, but did so at intermediate light levels and Follows 2016). Their generalist metabolic strategy (Appendix S1: Fig. S5). These results highlight the com- also makes them more likely to dominate late-succes- plexities of real mixotroph physiology, wherein different sional communities (i.e., late-season, stratified water col- taxa may derive different forms and degrees of benefit umns) where nutrients have been depleted (Stickney from ingesting prey (Mitra et al. 2016, Lie et al. 2018). et al. 1999, Mitra et al. 2014). Non-constitutive mixo- Mixotrophs are also known to adjust their metabolic trophs (which obtain their photosynthetic abilities from strategy as a function of environmental conditions: In their prey) may also coexist with stronger competitors, addition to photoacclimation responses to different light sometimes creating cyclic, bloom dynamics with pulsed levels, mixotrophs may downregulate photosynthetic biogeochemical impacts (Moeller et al. 2016); in con- machinery in the presence of high prey abundances (San- trast to the constitutive mixotrophs in our Ochromonas– ders et al. 1990, Holen 1999, Wilken et al. 2013, 2014b). Micromonas system, however, these taxa cannot compet- Such a response would cause Ochromonas to function itively exclude their prey because they rely on them as a more like a predator than a competitor, potentially alter- source of photosynthetic machinery (Moeller et al. ing dynamic outcomes, and/or accelerating extirpation 2016). of the prey. However, though we did not quantify cellu- In conclusion, by functioning as intraguild predators, lar chlorophyll-a content or other photophysiological mixotrophs that eat phytoplankton can coexist with parameters, we observed no differences in either per-cell these phytoplankton, even when the mixotrophs are the absorption coefficients or red fluorescence (a proxy for weaker competitor for the single limiting resource of photosynthetic pigment content, measured by the flow light. Our results provide further evidence that, across Xxxxx 2019 MIXOTROPHY IN PLANKTONIC COMMUNITIES Article e02874; page 13 aquatic and terrestrial systems, from microscopic to Flynn, K. J., and P. J. Hansen. 2013. Cutting the canopy to macroscopic organisms, intraguild predator persistence defeat the “Selfish Gene”: conflicting selection pressures for is mediated by resource supply levels and the strength of the integration of phototrophy in mixotrophic protists. 164:811–823. species interactions. These observations are consistent Frost, B. W. 1972. Effects of size and concentration of food par- ’ with the omnipresence of mixotrophs in the world s mar- ticles on the feeding behavior of the marine planktonic cope- ine and freshwater systems. pod Calanus pacificus. Limnology and Oceanography 17:805–815. ACKNOWLEDGMENTS Furuya, K., and R. Marumo. 1983. The structure of the phyto- HVM and MGN designed the model. HVM and MDJ plankton community in the subsurface chlorophyll maxima designed the experimental test system. HVM performed the in the western North Pacific Ocean. Journal of Plankton – model analysis, conducted the experiments, and analyzed the Research 5:393 406. data. All authors wrote the paper. We thank Susanne Wilken for Hammer, A. C., and J. W. Pitchford. 2005. The role of mixotro- generously providing axenic CCMP 2951 and 1393 cultures for phy in plankton bloom dynamics, and the consequences for – our use. R. Germain, S. Louca, G. Owens, N. Sharp, P. Thomp- productivity. ICES Journal of Marine Science 62:833 840. son, and J. Yoder provided valuable feedback on figure design. Hartmann, M., C. Grob, G. A. Tarran, A. P. Martin, P. H. Bur- We also thank J. Bronstein, S. Diehl, J. Huisman, C. Klausmeier, kill, D. J. Scanlan, and M. V. Zubkov. 2012. Mixotrophic and four anonymous reviewers for comments on earlier versions basis of Atlantic oligotrophic ecosystems. Proceedings of the of this manuscript. HVM was supported by a United States National Academy of Sciences of the United States of Amer- – National Science Foundation Postdoctoral Research Fellowship ica 109:5756 5760. in Biology (Grant DBI-1401332) and a University of British Heinbokel, J. F. 1978. Studies on the functional role of tintin- Columbia Research Centre Postdoctoral Fellow- nids in the Southern California Bight. I. Grazing and growth – ship. This material is based upon work supported by the rates in laboratory cultures. Marine Biology 47:177 189. National Science Foundation under Grants OCE-1655686 and Herzig, R., and Z. Dubinsky. 1992. Photoacclimation, photo- OCE-1436169, by a grant from the Simons Foundation/SFARI synthesis, and growth in phytoplankton. Israel Journal of – (561126, HMS), and by the Woods Hole Oceanographic Institu- Botany 41:199 212. tion’s Investment in Science Program. Research was also spon- Holen, D. A. 1999. Effects of prey abundance and light intensity sored by the U.S. Army Research Office and was accomplished on the mixotrophic chrysophyte Poterioochromonas mal- under Cooperative Agreement Number W911NF-19-2-0026 for hamensis from a mesotrophic lake. Freshwater Biology – the Institute for Collaborative Biotechnologies. 42:445 455. Holling, C. S. 1959. Some characteristics of simple types of pre- dation and . The Canadian Entomologist 91:385– 398. LITERATURE CITED Holt, R. D., and G. A. Polis. 1997. A theoretical framework for Adolf, J. E., D. Stoecker, and L. W. Harding Jr. 2006. The bal- intraguild predation. American Naturalist 149:745–764. ance of autotrophy and heterotrophy during mixotrophic Huisman, J. 1999. Population dynamics of light-limited phyto- growth of Karlodinium micrum (Dinophyceae). Journal of plankton: microcosm experiments. Ecology 80:202–210. Plankton Research 28:737–751. Huisman, J., and F. J. Weissing. 1994. Light-limited growth and Arim, M., and P. A. Marquet. 2004. Intraguild predation: a competition for light in well-mixed aquatic environments: an widespread interaction related to species biology. Ecology elementary model. Ecology 75:507–520. Letters 7:557–564. Hutchinson, G. E. 1941. Ecological aspects of succession in nat- Boraas, M. E., D. B. Seale, and D. Holen. 1992. Predatory ural populations. American Naturalist 75:406–418. behavior of Ochromonas analyzed with video microscopy. Hutchinson, G. E. 1961. The . Ameri- Archiv Hydrobiologie 123:459–468. can Naturalist 95:137–145. Borer, E. T., C. J. Briggs, W. W. Murdoch, and S. L. Swarbrick. Jeong, H. J., and M. I. Latz. 1994. Growth and grazing rates of 2003. Testing intraguild predation theory in a field system: the heterotrophic Protoperidinium spp. on red Does numerical dominance shift along a gradient of produc- tide dinoflagellates. Marine Ecology Progress Series 106:173– tivity? Ecology Letters 6:929–935. 185. Burkholder, J. M., P. M. Glibert, and H. M. Skelton. 2008. Jost, C., C. A. Lawrence, F. Campolongo, W. van de Bund, S. Mixotrophy, a major mode of nutrition for harmful algal spe- Hill, and D. L. DeAngelis. 2004. The effects of mixotrophy cies in eutrophic waters. Harmful Algae 8:77–93. on the stability and dynamics of a simple planktonic food Cottrell, M. T., and C. A. Suttle. 1991. Wide-spread occurrence web model. Theoretical Population Biology 66:37–51. and clonal variation in viruses which cause lysis of a cos- Katechakis, A., and H. Stibor. 2006. The mixotroph Ochromo- mopolitan, eukaryotic marine phytoplankter Micromonas nas tuberculata may invade and suppress specialist phago- pusilla. Marine Ecology Progress Series 78:1–9. and phototroph plankton communities depending on nutrient Crane, K. W., and J. P. Grover. 2010. Coexistence of mixo- conditions. Oecologia 148:692–701. trophs, autotrophs, and heterotrophs in planktonic microbial Keller, M. D., R. C. Selvin, W. Claus, and R. R. L. Guillard. communities. Journal of Theoretical Biology 262:517–527. 1987. Media for the culture of oceanic ultraphytoplankton. Cropp, R., and J. Norbury. 2015. Mixotrophy: the missing link Journal of Phycology 23:633–638. in -resource–based . Lie, A. A.-Y., Z. Liu, R. Terrado, A. O. Tatters, K. B. Heidel- 8:245–260. berg, and D. A. Caron. 2018. A tale of two mixotrophic Diehl, S., and M. Feissel. 2001. Intraguild prey suffer from chrysophytes: insights into the metabolisms of two Ochromo- enrichment of their resources: a microcosm experiment with nas species (Chrysophyceae) through a comparison of gene . Ecology 82:2977–2983. expression. PLoS ONE 13:e0192439-20. Fedriani, J. M., T. K. Fuller, R. M. Sauvajot, and E. C. York. Maat, D. S., K. J. Crawfurd, K. R. Timmermans, and C. P. D. 2000. Competition and intraguild predation among three Brussaard. 2014. Elevated CO2 and phosphate limitation sympatric . Oecologia 125:258–270. favor Micromonas pusilla through stimulated growth and Article e02874; page 14 HOLLY V. MOELLER ET AL. Ecology, Vol. xx, No. xx

reduced viral impact. Applied and Environmental Microbiol- mixotrophic chrysophyte Poterioochromonas malhamensis. ogy 80:3119–3127. 19:97–109. McKie-Krisberg, Z. M., and R. W. Sanders. 2014. Sanders, R. W., D. A. Caron, J. M. Davidson, M. R. Dennett, Phagotrophy by the picoeukaryotic green alga Micromo- and D. M. Moran. 2001. Nutrient acquisition and population nas: implications for Arctic Oceans. The ISME Journal growth of a mixotrophic alga in axenic and bacterized cul- 8:1953–1961. tures. Microbial Ecology 42:513–523. Mitra, A., et al. 2014. The role of mixotrophic protists in the Skovgaard, A. 1996. Mixotrophy in Fragilidium subglobosum biological carbon pump. Biogeosciences 11:995–1005. (Dinophyceae):growth and grazing responses as functions of Mitra, A., et al. 2016. Defining planktonic protist functional light intensity. Marine Ecology Progress Series 143:247–253. groups on mechanisms for energy and nutrient acquisition: Skovgaard, A., P. J. Hansen, and D. K. Stoecker. 2000. Physiol- incorporation of diverse mixotrophic strategies. Protist ogy of the mixotrophic Fragilidium subglobosum. 167:106–120. I. Effects of phagotrophy and irradiance on photosynthesis and Moeller, H. V., E. Peltomaa, M. D. Johnson, and M. G. Neu- carbon content. Marine Ecology Progress Series 201:129–136. bert. 2016. Acquired phototrophy stabilises coexistence and Stickney, H. L., R. R. Hood, and D. K. Stoecker. 1999. The shapes intrinsic dynamics of an intraguild predator and its impact of mixotrophy on planktonic marine ecosystems. Eco- prey. Ecology Letters 19:393–402. logical Modelling 125:203–230. Moore, L. R., A. F. Post, G. Rocap, and S. W. Chisholm. 2002. Stomp, M., J. Huisman, F. de Jongh, A. J. Veraart, D. Gerla, Utilization of different nitrogen sources by the marine M. Rijkeboer, B. W. Ibelings, U. I. A. Wollenzien, and L. J. cyanobacteria Prochlorococcus and Synechococcus. Limnol- Stal. 2004. Adaptive divergence in pigment composition pro- ogy and Oceanography 47:989–996. motes phytoplankton biodiversity. Nature 432:104–107. Morin, P. 1999. Productivity, intraguild predation, and popula- Strom, S. L. 2001. Light-aided , grazing and growth in tion dynamics in experimental food webs. Ecology 80:752– herbivorous protists. Aquatic Microbial Ecology 23:253–261. 760. Thingstad, T. F., H. Havskum, K. Garde, and B. Riemann. Mylius, S. D., K. Klumpers, A. M. de Roos, and L. Persson. 1996. On the strategy of “eating your competitor”: a mathe- 2001. Impact of intraguild predation and stage structure on matical analysis of algal mixotrophy. Ecology 77:2108–2118. simple communities along a productivity gradient. American Thompson, P. A., P. J. Harrison, and J. S. Parslow. 1991. Influence Naturalist 158:259–276. of irradiance on cell volume and carbon quota for ten species Not, F., M. Latasa, D. Marie, T. Cariou, D. Vaulot, and N. of marine phytoplankton. Journal of Phycology 27:351–360. Simon. 2004. A single species, Micromonas pusilla (Prasino- Tilman, D. 1977. Resource competition between plankton phyceae), dominates the eukaryotic in the algae: an experimental and theoretical approach. Ecology Western English Channel. Applied and Environmental Micro- 58:338–348. biology 70:4064–4072. Tilman, D. 1990. Constraints and tradeoffs: toward a predictive Paine, R. T. 1969. A note on trophic complexity and community theory of competition and succession. Oikos 58:3–15. stability. American Naturalist 103:91–93. Tittel, J., V. Bissinger, B. Zippel, U. Gaedke, E. Bell, A. Lorke, Passarge, J., S. Hol, M. Escher, and J. Huisman. 2006. Competi- and N. Kamjunke. 2003. Mixotrophs combine resource use to tion for nutrients and light: stable coexistence, alternative outcompete specialists: implications for aquatic food webs. stable states, or competitive exclusion? Ecological Mono- Proceedings of the National Academy of Sciences of the Uni- graphs 76:57–72. ted States of America 100:12776–12781. Polis, G. A., and R. D. Holt. 1992. Intraguild predation: the Van Baren, M. J., et al. 2016. Evidence-based green algal geno- dynamics of complex trophic interactions. Trends in Ecology mics reveals marine diversity and ancestral characteristics of and Evolution 7:151–154. land . BMC Genomics 17:267. Polis, G. A., C. A. Myers, and R. D. Holt. 1989. The ecology Verity, P. G., C. Y. Robertson, C. R. Tronzo, M. G. Andrews, J. and evolution of intraguild predation: potential competitors R. Nelson, and M. E. Sieracki. 1992. Relationships between that eat each other. Annual Review of Ecology and Systemat- cell volume and the carbon and nitrogen content of marine ics 20:297–330. photosyn thetic nanoplankton. Limnology and Oceanogra- Price, J. E., and P. J. Morin. 2004. Colonization history determi- phy 37:1434–1446. nes alternate community states in a of intraguild Ward, B. A., and M. J. Follows. 2016. Marine mixotrophy predators. Ecology 85:1017–1028. increases trophic transfer efficiency, mean organism size, and Ptacnik, R., U. Sommer, T. Hansen, and V. Martens. 2004. vertical carbon flux. Proceedings of the National Academy of Effects of microzooplankton and mixotrophy in an experi- Sciences of the United States of America 113:2958–2963. mental planktonic food web. Limnology and Oceanography Wilken, S., J. Huisman, S. Naus-Wiezer, and E. Van Donk. 49:1435–1445. 2013. Mixotrophic organisms become more heterotrophic Ptacnik, R., et al. 2016. A light-induced shortcut in the plank- with rising temperature. Ecology Letters 16:225–233. tonic . Scientific Reports 6:29286. Wilken, S., J. M. H. Verspagen, S. Naus-Wiezer, E. Van Donk, Raven, J. 1997. Phagotrophy in phototrophs. Limnology and and J. Huisman. 2014a. Biological control of toxic cyanobac- Oceanography 42:198–205. teria by mixotrophic predators: an experimental test of intra- Rothhaupt, K. O. 1996a. Laboratory experiments with a mixo- predation theory. Ecological Applications 24:1235–1249. trophic chrysophyte and obligately phagotrophic and pho- Wilken, S., J. M. H. Verspagen, S. Naus-Wiezer, E. Van Donk, totrophic competitors. Ecology 77:716–724. and J. Huisman. 2014b. Comparison of predator-prey interac- Rothhaupt, K. O. 1996b. Utilization of substitutable carbon tions with and without intraguild predation by manipulation and phosphorus sources by the mixotrophic chrysophyte of the nitrogen source. Oikos 123:423–432. Ochromonas sp. Ecology 77:706–715. Wissinger, S., and J. McGrady. 1993. Intraguild predation and Sanders, R. W., and R. J. Gast. 2011. Bacterivory by pho- competition between larval dragonflies: direct and indirect totrophic picoplankton and nanoplankton in Arctic waters. effects on shared prey. Ecology 74:207–218. FEMS Ecology 82:242–253. Worden, A. Z., et al. 2009. Green evolution and dynamic adap- Sanders, R. W., K. G. Porter, and D. A. Caron. 1990. Relation- tations revealed by genomes of the marine picoeukaryotes ship between phototrophy and phagotrophy in the Micromonas. Science 324:268–272. Xxxxx 2019 MIXOTROPHY IN PLANKTONIC COMMUNITIES Article e02874; page 15

SUPPORTING INFORMATION Additional supporting information may be found in the online version of this article at http://onlinelibrary.wiley.com/doi/ 10.1002/ecy.2874/suppinfo

DATA AVAILABILITY Model code is available from Zenodo: https://doi.org/10.5281/zenodo.3350438