Competition and succession among coral endosymbionts

Abstract Host species often support a genetically diverse guild of symbionts, the identity and performance of which can determine holobiont fitness under particular environmental conditions. These symbiont communities are structured by a complex set of potential interactions, both positive and negative, between the host and symbionts and among symbionts. In reef‐building corals, stable associations with specific symbiont species are common, and we hypothesize that this is partly due to ecological mechanisms, such as succession and competition, which drive patterns of symbiont winnowing in the initial colonization of new generations of coral recruits. We tested this hypothesis using the experimental framework of the de Wit replacement series and found that competitive interactions occurred among symbionts which were characterized by unique ecological strategies. Aposymbiotic octocoral recruits within high‐ and low‐light environments were inoculated with one of three Symbiodiniaceae species as monocultures or with cross‐paired mixtures, and we tracked symbiont uptake using quantitative genetic assays. Priority effects, in which early colonizers excluded competitive dominants, were evidenced under low light, but these early opportunistic species were later succeeded by competitive dominants. Under high light, a more consistent competitive hierarchy was established in which competitive dominants outgrew and limited the abundance of others. These findings provide insight into mechanisms of microbial community organization and symbiosis breakdown and recovery. Furthermore, transitions in competitive outcomes across spatial and temporal environmental variation may improve lifetime host fitness.

Within multispecies symbioses, dynamic networks of direct and indirect effects, well beyond traditional pairwise interactions (Stanton, 2003), determine symbiont community structure via interactions among multiple co-occurring symbionts, their host, and the environment (Palmer, Stanton, & Young, 2003). Environmental conditions may alter competitive hierarchies among symbionts as different resources become more or less limiting, and both host and symbionts may modify the in-hospite environment in ways that can either facilitate or exclude additional symbiont types (Palmer et al., 2002;Wangpraseurt, Larkum, Ralph, & Kühl, 2012). During initial symbiont uptake in horizontally transmitting hosts (e.g., the majority of corals), priority effects, in which early arrivals are less prone to displacement, can change the trajectory of symbiont community succession and/or allow for species coexistence (Fukami, 2015;Fukami et al., 2010;Halliday, Umbanhowar, & Mitchell, 2017;Palmer et al., 2002).
In this study, we focused on the role of competition and succession in the initial establishment of symbiont communities in newly settled coral recruits. We adapted the framework and basic expectations of the de Wit replacement series design (De Wit, 1960;Harper, 1967) (Figure 1) to evaluate competition, coexistence, and species turnover within newly available host habitat. We offered three Symbiodiniaceae species as monocultures and as three cross-paired mixtures (0.5:0.5 ratio) to aposymbiotic octocoral recruits (Briareum asbestinum) and used quantitative genetic assays to determine the presence and abundance of each species within each coral recruit.
We considered three models of competitive opportunistic niche exploitation: (a) competitive exclusion (one symbiont excludes another from entering into symbiosis at detectable levels) which would favor the first symbiont to enter symbiosis; (b) competitive dominance (in which one symbiont reduces the abundance of a co-occurring symbiont) which would favor fast proliferation; and (c) a null model (no competition), in which symbiont uptake would follow the availability of each type in the environment regardless of whether additional symbiont types were present (Figure 1). We also tested whether these interactions are modulated by light levels as this is a known environmental factor that influences Symbiodiniaceae distributions in nature (Kemp, Fitt, & Schmidt, 2008;Rowan et al., 1997).

| ME THODS
Our study species, B. asbestinum, is an abundant octocoral species throughout the Caribbean (Bayer, 1961) and which, following settlement, can simultaneously host as many as six different Symbiodiniaceae species from four deeply divergent genera (Poland F I G U R E 1 de Wit replacement series model for symbiont competition. Two potential symbiont species are offered at ratios of 1:0, 0.5:0.5, or 0:1. In-hospite densities of each symbiont species measured within monocultures (circles) are used to model expected in-hospite densities in the absence of competition (dashed lines). Within duoculture, measured in-hospite densities that fall near (gray square) or significantly below (black square) expected values signify the absence or presence of competition, respectively et al., 2013). We focused on the early weeks of symbiont uptake in B. asbestinum, during which host selection mechanisms appear to be weak, as this provided an opportunity for greater insight into alternative factors, specifically competition among symbionts, in influencing symbiont community structure. Surface brooded B. asbestinum larvae were collected from more than 10 adult colonies in a single day in the Middle Florida Keys (24º49′38″N,80º48′50″W) which appeared to be approximately 2 days old (B. asbestinum larvae remain clumped on branches for 3-5 days; Brazeau & Lasker 1990).
Larvae were transported to the Keys Marine Laboratory where they were washed several times in 0.45 µm filtered seawater (FSW), and maintained in FSW. Sun-dried, dead gorgonian branches were provided as a settlement substrate (Coffroth et al., 2006;Poland et al., 2013), onto which larvae attached and metamorphosed into single   (Guillard 1975), at ~27°C, under fluorescent lighting on a 14:10 hr light:dark regime. New batch cultures were regularly restarted to ensure growing and swimming cell populations. Each batch culture was sampled for chloroplast genotyping (Santos et al., 2003) to ensure purity. The use of isoclonal Symbiodiniaceae cultures allowed controlled manipulation of the number of cells of each species made available to the initially aposymbiotic hosts. Species within the ITS2-type B1 group dominate 1-to 2-year-old B. asbestinum juveniles (Poland et al., 2013); the B. minutum culture used in this study has been found in newly settled field recruits, but it is  Thornhill, LaJeunesse, Kemp, Fitt, & Schmidt, 2006) and have been found in newly settled B. asbestinum in the field (Coffroth et al., 2006;Poland et al., 2013).
We used a single, cross-factorial experimental design nesting light environment within each symbiont treatment, which included either one or two symbiont species. Symbiont treatments included:
Quantitative PCR (qPCR) was used to quantify the abundance of symbiont cells using genus-specific assays (Symbiodinium, prev. Clade A (Winter, 2017); Breviolum, prev. Clade B (Cunning, Vaughan, et al., 2015); Durusdinium, prev. Clade D (Cunning & Baker, 2013)). All assays were pretested to confirm the applicability of these assays for the three Symbiodiniaceae species employed in this experiment. A set of control recruits (n = 5), which were not inoculated with any symbiont cultures, were also tested for the presence of symbionts. DNA extracted from each recruit was assayed with two technical replicates of each genus-specific primer set. Reaction volumes were 10μL with 5μL Taqman Genotyping Master Mix and 1μL genomic DNA template.
Assays were optimized for each target including: Symbiodinium, prev.
To convert C T values to cell numbers per recruit, standard curves of DNA from known numbers of cells (2,000, 4,000, 16,000, 32,000, and 64,000 cells/sample) of all three cultured Symbiodiniaceae species (extracted using identical methods and volumes as experimental recruits) were amplified in duplicate on each qPCR plate. The known cell numbers for all standard curve amplifications were log-transformed and modeled as a function of C T value and target genus using a linear mixed model with random slopes and intercepts for each qPCR plate. The fitted model was then used to calculate the number of cells of each phylotype in each unknown sample using C T value, target genus, and qPCR plate as predictors (all data and analysis code available at https :// github.com/shelb y26/Mixed-Uptake). We also excluded any cases in which noninoculate symbionts were detected.

| Total cell uptake
To examine patterns of symbiont uptake for each of the symbiont species when offered individually, we compared the total number of cells per recruit with a two-way ANOVA (fixed factors: symbiont treatment, light, and their interaction) at each time point. To conform with the assumptions of ANOVA, cell numbers were first log transformed to fit a normal distribution. When significant differences were detected, a Tukey's HSD post hoc test was performed limited to a priori, within factor comparisons. All analyses were run in R (R Core Team 2015).
The value for total cells per recruit within the mixed treatment was incorporated into separate analyses, see below.

| Competitive exclusion
We first tested the ability of each symbiont species to exclude others from entering the symbiosis (i.e., competitive exclusion). Within the mixed inoculation treatments, we pooled data from the two time points and analyzed the frequency with which each symbiont type was present or absent (below the level of detection by qPCR) in B. asbestinum recruits. These frequencies were tested in a Pearson's chi-squared test with Yate's correction for continuity.

| Competitive dominance
To test for deviations from the expected values of the de Wit null model in each mixed-uptake combination, the number of cells of a  (Lenth, 2016) was then used to test for significant differences between expected and observed number of cells for each symbiont type within each light level.

| Single-infection treatments
All symbiont species were taken up readily when offered individually in both the high-and low-light treatments (Figure 3). At the 6 week time point, polyps harbored 52%-74% fewer symbionts in low-light treatments relative to high light (F 1,50 = 7.75, p = .008).
There was also an effect at 6 weeks of symbiont species on cell den-  Figure 3). No symbionts were detected in the control recruits. We did not test for an effect of host tissue mass qPCR efficiency; however, we measured similar and increasing symbiont densities through time.

| Competitive exclusion
The detection of only a single symbiont type where two were offered occurred in 26% of recruits sampled from the high-light treatment and 44% of recruits in the low-light treatment. Within the low-light treatment, we found that B. minutum was absent (or below the level of detection by qPCR) more frequently than expected when occurring with either S. microadriaticum or D. trenchii (χ 2 test; p < .05;

| D ISCUSS I ON
We were able to establish a general competitive hierarchy in which In isolation, all Symbiodiniaceae species were shown to proliferate more slowly under lower irradiance (Figure 3), supporting a high similarity in requirements among species (at least for light, which is the primary environmental gradient for photosynthetic organisms).
In competition, low light frequently led to the exclusion of B. minutum from polyps despite simultaneous environmental availability of species pairs (Table 2; Figure 4). While competitive dominance can Note: For each mixed symbiont and light treatment, the presence and absence of each symbiont within a sampled Briareum asbestinum recruit were recorded. Samples at 6 and 8 weeks were pooled. Symbiodinium microadriaticum (S. mic.), Breviolum minutum (B. min), Durusdinium trenchii (D. tren.). These frequencies were analyzed with a chi-square test with Yate's correction. Bold values with asterisk indicate that a particular symbiont was excluded significantly more than expected by random chance alone p<0.05.  Figure 1). Significant deviations are noted with asterisks, and the letters and arrows in the upper left of the graph indicate direction of deviation within the given symbiont type. Error bars show standard error of the mean & Koike, 2014). Subsequent priority effects then occur where early arriving species have a large impact on that niche (Palmer et al., 2002), and where late arrivals are highly sensitive to niche availability (Fukami, Beaumont, Zhang, & Rainey, 2007). Indeed, the cell size of both S. microadriaticum and D. trenchii is large relative to B. minutum (Biquand et al., 2017;LaJeunesse, 2001;LaJeunesse, Lambert, Andersen, Coffroth, & Galbraith, 2005;Suggett, Goyen, & Evenhuis, 2015) which could more rapidly lead to light limitation via shading (Cunning & Baker, 2013) as these larger cells populate host tissues.
Furthermore, larger Symbiodiniaceae cells tend to have an increased ability for light-harvesting which may lead to a positive feedback in their competitive advantage over smaller cells (Suggett et al., 2015).

The long-term trajectory of B. asbestinum juveniles in favor of
Breviolum spp. has been demonstrated in both the laboratory and field (Poland & Coffroth, 2017Poland et al., 2013). Indeed, numerous coral species establish predictable symbiont associations over time, regardless of the initial composition of symbiont communities (Abrego et al., 2009;Little et al., 2004;McIlroy & Coffroth, 2017;Poland & Coffroth, 2017;Poland et al., 2013;Quigley, Willis, & Bay, 2016). However, even short-term competition and succession among symbionts may have important consequences for host fitness. While we did not quantify the impact of symbiont composition on hosts in this study, a laboratory study by (Poland & Coffroth, 2019) found that, by 3 months of age (1 month beyond our own study), B. asbestinum hosting Symbiodinium sp. or mixed communities of Symbiodinium sp. and Breviolum spp. had slower growth (i.e., polyp budding rates) and higher mortality relative to those hosting only Breviolum. In the field, selection for fast succession to optimal symbionts may be even more pronounced, particularly for juveniles corals that have an inverse relationship between size and mortality (Edmunds & Gates, 2004). In fact, this seems to be the case with B. asbestinum where the majority of field-reared polyps are dominated by Breviolum by three months (Poland et al., 2013). In symbioses, infectivity and high rates of in-hospite proliferation are generally associated with parasitism because of their demand on a shared pool of nutritional resources (Baker et al., 2018;Sachs & Wilcox, 2006).
However, balancing selection for both competitive (i.e., self-promoting) and mutualistic (i.e., promoting host growth and fitness) traits may ultimately lead to evolution of predictable and beneficial hostsymbiont associations even in the absence of host control.
Despite the global ubiquity of members of Durusdinium at very low relative abundance in corals (Silverstein, Correa, & Baker, 2012;Tong et al., 2017), the poor competitive ability of D. trenchii shown here may contribute to its uncommonness as a dominant symbiont, except following bleaching (the stress-induced loss of symbionts from the host). Furthermore, the fact that D. trenchii is not particularly competitive under the conditions studied here may promote the reversion to alternative dominant symbiont types following recovery (Jones et al., 2008;Thornhill et al., 2006) Figure 4). Competitive outcomes were tested at both 6 weeks and 8 weeks following initial inoculations. Symbiodinium microadriaticum (S. mic.), Breviolum minutum (B. min), Durusdinium trenchii (D. tren.) change depending on environmental condition (Cunning, Gillette, Capo, Galvez, & Baker, 2014). In this way, competitive outcomes can also underpin the complementary or even synergistic benefits of multispecies mutualisms on lifetime coral fitness (Palmer et al., 2010).
We were able to demonstrate that competition among symbionts influences the colonization of new hosts, but we found that our competitive hierarchies were not generalizable through time with variable outcomes at 6 and 8 weeks. One explanation is that a host is not a static habitat. Changes in the host environment, distribution of symbiont cells among tissues, and/or nutrient sharing can occur as hosts grow (Lecointe, Domart-Coulon, Paris, & Meibom, 2016) which may alter the competitive hierarchy.
Furthermore, while the small size of recruits limited our ability to assess both symbiont genetics and host tissue mass simultaneously, differences in recruit growth among treatments may have fed-back into competitive outcomes. Symbionts also become more densely packed into host tissues overtime. The approximately 10fold increase in symbiont densities between 6 and 8 weeks across light treatments may allow for density-dependent effects on competitive abilities (Cunning, Vaughan, et al., 2015). Thus, long-term associations may favor those symbionts that compete well and es- Previously, models that compared hosts with single versus mixed symbiont assemblages have been used to understand coevolution (Gomulkiewicz, Nuismer, & Thompson, 2003;Hoeksema & Kummel, 2003), to better predict the effect of mutualists on host-enemy interactions (McKeon, Stier, McIlroy, & Bolker, 2012;Morris et al., 2007;Palmer et al., 2008), to predict spatial and environmental characters that promote mutualism function (Boza & Scheuring, 2004;Doebeli & Knowlton, 1998) and to understand host ontogeny (Palmer et al., 2010). The compact, easily replicated, coral-algal mutualism presents an excellent model system to further explore these phenomena and provide new perspectives on the consequences of diversity and flexibility in symbiosis in general. Future research, aided by techniques that can quantify absolute and relative abundances of specific symbionts in mixed associations (e.g., qPCR, high-throughput sequencing, and genetic tagging) and more directly demonstrate their effects on host fitness (e.g., physiology) will continue to reveal the ecology and evolution of diverse symbioses ubiquitous throughout the earth's ecosystems.

ACK N OWLED G M ENTS
We would like to thank R. Mellas, A. Kleuter, and the staff at Long Key Marine Laboratories for supporting fieldwork, H. Lasker for input on the data analyses, and N. Knowlton, T. Bonebrake, J. D.
Gaitan, and D.M. Baker for comments on the manuscript. Funding was provided by NSF OCE-09-26822 (MAC).

CO N FLI C T O F I NTE R E S T
All authors declare that there are no competing financial interests in relation to the work described.

AUTH O R CO NTR I B UTI O N S
SEM and MAC conceived the projects and executed experiments, RC and ACB developed genetic methodologies, SEM and RC ran the genetic assays and data analyses, all authors contributed to the writing of the manuscript and have approved the final version.

DATA AVA I L A B I L I T Y S TAT E M E N T
Upon publication, all data and analysis code will be available at https ://github.com/shelb y26/Mixed-Uptake.