Food nutrient availability affects epibiont prevalence and richness in natural Daphnia populations

The increased input of nutrients into biological systems has been shown to result in altered biotic interactions through changes in food availability. The aim of this study was to test for an association between phytoplankton nutrient content and epibiont variables in natural zooplankton populations. Via a field survey, we studied how a gradient in food quantity and quality impacted host population density and epibiont variables in Daphnia pulex. We found a significant decrease in epibiont prevalence and infracommunity richness, which could mainly be attributed to a changing phytoplankton N : P ratio (caused by P‐limitation). We performed a lab experiment in which we exposed Daphnia magna to different algal nutrient ratios and the epibionts detected in the field study. P‐limitation in the algae affected D. magna performance and resulted in similar trends of food quality effects in the epibiont variables. The experiment, however, also reflected subtle differences between different epibiont species.

Eutrophication is a persistent threat to biodiversity in natural ecosystems and especially in freshwater ecosystems (Jones and Brett 2014). Nitrogen (N)-and phosphorus (P)-rich pollutants are released in ponds and streams, strongly influencing ecosystem functioning (Gulati and van Donk 2002). Changes in these nutrients are particularly relevant when considering the effect of anthropogenic-induced nutrient inputs in natural ponds and biotic and abiotic interactions (Elser et al. 2010;Peñuelas et al. 2012;Frenken et al. 2017;Narr et al. 2019). Global N : P ratios in aquatic ecosystems have changed since the industrial revolution, mainly due to the disproportional increase in C-and N-concentration relative to P-concentration (Peñuelas et al. 2012). Changed nutrient supply rates to an aquatic ecosystem often result in important modifications of food quality for invertebrate consumers (Sterner and Elser 2002;van de Waal et al. 2010;Hessen et al. 2013). Zooplankton, for example, have been found to be negatively affected by both the relative limitation or excess of either N or P in their food resources (Boersma &Elser 2006, Zhou andDeclerck 2019). For instance, a high phytoplankton N : P ratio can serve as low quality food for zooplankton due to relative excess of nitrogen (Reyserhove et al. 2017a,b), or limitation by phosphorus (Andersen and Hessen 1991;Sterner et al. 1997;Acharya et al. 2004;Zhou et al. 2018).
We here focus on the effect of food quality on one type of biotic interaction, in particular the zooplankton-epibiont interaction. Epibionts are organisms that live attached to its host body surface and are known to affect the zooplankton host. Positive effects (Barea-Arco et al. 2001;Fernandez-Leborans 2010) as well as negative effects of epibionts have been documented (Burris and Dam 2014;Conde-Porcuna et al. 2014;Pauwels et al. 2014). Epibionts mainly use the host as a substratum for attachment, but can mechanically interfere and compete for nutrient intake with the host (Pauwels et al. 2014). Studies have shown that epibiont burden is positively dependent on host body size (Threlkeld and Willey 1993;Krasnov et al. 2004), and transmission rate is expected to be positively dependent on host population density (Decaestecker et al. 2005;Ebert 2005). Knowledge on the effect of environmental changes in N : P ratios on epibiont variables and which mechanisms are involved is lacking. Increases in N-or P-loading may not always have a detectable effect on biotic interactions because of the complexity of the ecosystem and the multitude of possible drives (Aalto et al. 2015). Other biotic factors such as predation pressure (Guttierez et al. 2016), interspecific competition , or abiotic factors such as temperature changes Brans et al. 2017;Hessen et al. 2017) may interact with studied variables in generating the detected associations (van de Waal et al. 2018).
We investigated the effect of a stoichiometric mismatch on the performance of Daphnia and demographic features of its epibionts. Daphnia is a widespread keystone genus in freshwater ecosystems as it provides a link between primary producers and the upper levels in the food chain (Ebert 2005;Lampert 2006). In particular, we studied how phytoplankton N : P ratio impacts Daphnia epibiont prevalence, association intensity, and species richness in natural Daphnia populations and in experimental Daphnia cohorts. We focus on this food quality variable as it is particularly relevant in the context of increasing N to P supply rates to aquatic ecosystems (Peñuelas et al. 2012(Peñuelas et al. , 2013 and as nutrient availability (both N and P) have been suggested to impact Daphniaendoparasite interactions (Aalto et al. 2014;Reyserhove et al. 2017a,b;Narr et al. 2019). We expect that the effect of stoichiometric imbalance in the elemental supply of the environment will have an effect on the species composition and relative abundance of the epibionts. Whether the epibionts are autotrophic, saprophytic or suspension feeders feeding on algae being plastic with respect to elemental composition. Given that epibionts feed on seston, it can be expected that they will be affected under a stoichiometric mismatch. However, some epibionts do not directly feed on algae and may depend on Daphnia food sources (and performance) as well.

Materials and methods
We performed a field survey in combination with a laboratory experiment to investigate the effect of food quality on epibiont variables. For the field survey 12 ponds that contained Daphnia pulex populations (Table S1) were selected out of a large sampling campaign. Between May and June 2013, a total of 81 shallow ponds covering the central region of Flanders (Belgium; see Data SI1) were sampled as part of the Belspo Interuniversity Attraction Pole project SPEEDY (Spatial and Environmental determinants in Eco-Evolutionary Dynamics) investigating the effect of urbanization on organismal traits (Brans et al. 2017). The selected ponds were shallow, fishless and had a surface area smaller than 1 ha. As the ponds were sampled once, we cannot generalize our results for one pond over the field season. The different ponds were sampled over the field season ad random.

Characterization of abiotic and phytoplankton variables of SPEEDY ponds
For 57 of the 81 sampled ponds, phytoplankton total N-concentration (P TN , % DW), phytoplankton total P-concentration (P TP, % DW) and phytoplankton C : N, C : P, and N : P (molar) ratios were determined as proxies for food quality (Data SI1 and Table S2). Chlorophyll a (Chl a) and phytoplankton total carbon concentration (P TC , % per phytoplankton dry weight [DW]) were determined as proxies for food quantity. Pond physical variables (temperature, oxygen, pH, and conductivity) and chemical variables (total phosphorus, total nitrogen, dissolved organic carbon, and suspended matter) were also quantified (Data SI1).
Zooplankton and epibiont sampling and variables in selected SPEEDY ponds D. pulex populations were sampled with a tube sampler at four different locations in the pond (in different quadrants from the limnetic part to the shore) in order to determine the population density and the mean body size. Forty liter were filtered over a 64 μm sieve and fixated with formalin to determine D. pulex population density. An additional sample was taken via a sweep net (64 μm mesh size) and stored on ice to randomly isolate 20-30 living female adults and check them for the presence of epibionts within 24 h of sampling. Epibionts were identified to taxa according to Decaestecker et al. (2005) and Ebert (2005) (see Table 1). Adult females were inspected using a light microscope with reflected (organisms were illuminated from the top against a dark background) and transparent light (from below through the transparent organisms). The prevalence of a particular epibiont species was defined as the number of infected D. pulex individuals with that epibiont species relative to the total number of screened D. pulex individuals for a given pond. Due to the high amount of zero prevalence samples for Colacium sp., and because Brachiounus rubens, Vorticella sp. and Amoebidium parasiticum prevalence roughly responded in a similar way to the changing food variables, the epibiont prevalence data was pooled into one single variable for all epibiont species. This "pooled epibiont prevalence" was determined as the number of infected D. pulex individuals with at least one epibiont species relative to the total number of screened individuals for a given pond. Epibiont infracommunity richness was defined as the number of epibiont species infecting one zooplankton individual. Association intensity of a particular epibiont was defined as the total number of epibiont individuals (number of propagules under the stereomicroscope) in or on a single infected D. pulex individual, excluding non-infected animals. We did not consider a pooled association intensity variable (i.e., a single value representing the association intensity of all epibiont individuals on one D. pulex individual) given that the variation in the number of spores or propagules among epibiont species was very large.

Laboratory Daphnia-epibiont experiment
In order to confirm the effect of N : P ratio on Daphnia epibionts from the field survey, we performed an additional experiment in which the N : P ratio of Chlamydomonas reinhardtii was experimentally manipulated when fed to infected Daphnia magna (KNO 15.04 clone isolated from a small pond near Knokke, at the Belgian coast 51 20 0 05.62 00 N, 03 20 0 53.63 00 E) cohorts.
Using single stage chemostats, we established three C. reinhardtii cultures with different molar N : P ratios: a low N : P culture (mean: 16; SD: 3.0), a medium N : P culture (mean: 27; SD: 2.4), and a high N : P culture (mean: 74; SD: 4.44). Differences in N : P ratio were obtained by varying the concentrations of phosphorus in the culture medium. The algal food cultures were maintained in three continuous cultures using 2 liter-single stage chemostats that were continuously fed with modified WC   medium at a temperature of 23 AE 1 C and a dilution rate of 0.33 d −1 . We varied concentrations of dissolved inorganic nutrients (KH 2 PO 4 and NaNO 3 ) between the chemostats to realize algae with different N : P ratios: medium of "High N : P" cultures contained 170 mg L −1 NaNO 3 and 2.613 mg L −1 KH 2 PO 4 ; medium of "medium N : P" cultures contained 170 mg L −1 NaNO 3 and 4.36 mg L −1 KH 2 PO 4 ; and medium of "low N : P" cultures contained 85 mg L −1 NaNO 3 and 8.71 mg L −1 KH 2 PO 4 . We first created 18 cohorts of 30 D. magna juveniles of 3-4 d old. Each multifactorial combination of N : P food and epibiont treatments were replicated independently three times (cohorts of three independent maternal lines). Three maternal lines were set up. Five individuals per maternal line were transferred in 500 mL filtered tap water and fed every other day with a saturating amount of Chlorella vulgaris. Medium was refreshed every week. Juveniles from the first brood were discarded. Twenty juveniles from the second brood per maternal line were transferred in 2 liter-jars of filtered tap water. Parental Daphnia were discarded after releasing their second brood. This process was repeated for every new batch of 2 nd brood juveniles up to three generations.  other day until the individuals were fed with pure C. reinhardtii.
Hereafter the maternal lines were maintained by feeding every other day with a saturating amount of C. reinhardtii. For practical reasons related to monitoring, we divided all units from each maternal line per treatment combination in two technical replicates (with each a cohort of 15 juveniles) before exposure to the epibionts.
The experimental design consisted of three food N : P levels and two epibiont treatments (i.e., a control and a mixture of A. parasiticum and Vorticella). For the epibiont treatment, we exposed experimental individuals to a mix of A. parasiticum and Vorticella sp. For this, Daphnia individuals infected with epibionts were collected from a pond in Kortrijk, Belgium (50 48 0 05.7 00 N 3 16 0 32.9 00 E). These individuals were squashed and pooled per replicate prior to use in the experiment. Daphnia individuals of the experiment were each exposed to the equivalent of 1/3 infected individual. We thus created three squashed samples and applied part of each of these samples to two technical replicates of each of the factorial combinations. From 0-3/4 d old: the daphnids were fed every other day with a saturating amount of C. reinhardtii. From 4/5-6/7 d old, the daphnids were not fed (start-up of the experiment and stimulation uptake epibionts). From 7/8 d old until the end experiment: the daphnids were fed every day with 1 mg C L −1 of C. reinhardtii. Daphnia life history traits and body size, epibiont association intensity and prevalence were examined on one of the two technical replicates on four time points: on day 13, day 15, day 20, and day 22. Epibiont prevalence and association intensity was determined by screening a random subsample of five individuals per cohort. Individuals were individually screened using a light microscope with reflected and transparent light. The prevalence of each epibiont species was defined as the number of infected D. magna individuals relative to the five inspected individuals per cohort. The association intensity of each epibiont species was defined as the number of individuals of the respective epibiont species present on the carapace of the host individual. Body size was determined by taking a picture of each screened individual and analyzing body size using the ImageJ software. Body size was defined as the distance between the head and the base of the tail. Survival was measured on six time points: on day 5, day 8, day 13, day 15, day 20, and day 22. Survival was defined as the percentage of surviving individuals per cohort relative to the number of individuals on day 1.

Statistical analyses epibiont field data
To determine the relevance of the phytoplankton food quality and food quantity variables in the larger set of biotic and abiotic pond variables (total nitrogen concentration, total phosphorus concentration, dissolved organic carbon concentration, suspended matter, pH, conductivity, temperature and oxygen level), and to reduce the amount of independent variables in the statistical analyses, a principal component analysis (PCA) was performed on the dataset of SPEEDY ponds (n = 57, originally there were 81 ponds in the SPEEDY data set, but only 57 contained all variable measurements needed for the PCA). For this analysis, all variables were standardized (scaled), and temperature, TP, TN, SM, DOC, Chl a, N : P ratio, and P TN were log transformed to fulfill the assumptions of normality. We included the larger pond data set (57 ponds rather than the 12 selected D. pulex ponds) as this allowed us to carry out a more powerful analysis. This pond data set was inclusive the 12 ponds for which we could analyze epibiont prevalence and intensity on D. pulex (Table S1).
Epibiont prevalence, association intensity and infracommunity richness (log + 1 transformed) were analyzed by generalized linear mixed models (GLMMs) with a binomial (prevalence) and Poisson (intensity and richness) distribution respectively. We accounted for overdispersion by including the observation-level effect as a random effect (as in Harrison 2014). In the association intensity GLMM, values for each epibiont species were used and we accounted for epibiont identity (by including "epibiont species identity" as an extra explanatory variable). Multiple counts for association intensity per zooplankton individual per pond were available, as multiple epibionts can infect one host individual. Non-infected individuals were excluded from these association intensity analyses given the redundancy with prevalence. Host individual was nested in pond and pond identity was considered as a random factor. To account for interdependence, host individual was nested in pond and identity was included as a random effect. All statistical tests were performed in R 3.6.1 (R Core Team).

Statistical analyses Daphnia-epibiont experiment
Body size and survival was analyzed to confirm the effect of the N : P treatments on Daphnia performance. Body size was analyzed using a GLM assuming a Gaussian distribution of the data. Survival was analyzed for the last time point with a Logrank (Mantel-Cox) test. All analyses on association intensity and prevalence were performed on each of the four time points independently, and for A. parasiticum separately, Vorticella sp. separately and both epibionts pooled. Association intensity was analyzed using a GLM assuming a Poisson distribution of the data and accounted for overdispersion when the observed residual deviance was higher than the degrees of freedom. Prevalence was analyzed using a GLM assuming a Binomial distribution of the data and accounted for overdispersion when the observed residual deviance was higher than the degrees of freedom. For the association intensity, non-associated individuals were left out of the analysis given the redundancy with prevalence. Analyses on association intensity for Vorticella sp. were not possible due to too few data points. In all these analyses, we evaluated the effect of the factor 'maternal line' by comparing AIC values of models with and without inclusion of this factor. Maternal line was included if the AIC value of the corresponding model was more than two units smaller than Table 4. Regression analysis on the impact of phytoplankton N : P ratio, total nitrogen concentration (P TN ) and D. pulex population density on epibiont variables (prevalence, infracommunity IF richness and association intensity). z, Wald's z value; p, p value. Significant p values are indicated in italics. IF richness: random effect = pond identity; variance = 0.0148; SD = 0.1217. Association intensity: random effect = pond identity; variance = 2.172; SD = 1.474. Epibiont species identity: chi-squared value: 91.26, p < 0.001. the model without this factor. All statistical tests were performed in R 3.6.1 (R Core Team).

Results
Epibiont sampling in the SPEEDY field design Four epibionts (Vorticella sp., B. rubens, A. parasiticum, and Colacium sp.) were detected in the D. pulex populations (Table 1 and Fig. 1). Epibionts were present in 11 of the 12 sampled ponds.
In the PCA-analysis on the studied ponds, the percentage of explained variation of PCA1 and PCA2 was 25.73% and 15.75%. Oxygen, total phosphorus and total nitrogen were the most pronounced variables according to the PCA1-axis, while Chl a correlated most with the PCA2-axis ( Fig. 2; Table 2). PCA1-axis revealed a significant negative effect on epibiont prevalence and infracommunity richness, PCA2-axis (mainly food quantity) did not show any effect ( Fig. 3; Table 3).
Regression analysis on the studied ponds (correlation on log-transformed data between N and N : P: r = 0.055, df = 21, and p = 0.8 and between P and N : P: r = −0.62, df = 21, and p = 0.002, n = 12) showed that variation in the seston N : P ratio was more strongly driven by P TP than by P TN indicating that seston N : P ratio mainly reflects P-limitation of the food resource rather than excess N.
A high phytoplankton N-concentration and N : P ratio (P-limitation) was significantly associated with a low epibiont prevalence (Fig. 4a,c; Table 4). High values of the seston N-concentration and N : P ratio (P-limitation) were associated with a low epibiont infracommunity richness (Fig. 4b,d; Table 4). A high D. pulex population density was significantly associated with a high epibiont prevalence ( Fig. 4e; Table 4) and more epibiont species per host individual ( Fig. 4f; Table 4). No significant effects were found on epibiont association intensity (Table 4). Food N : P ratio and D. pulex population density were uncorrelated.
Laboratory Daphnia-epibiont experiment A high food N : P ratio (P-limitation) was associated with a smaller Daphnia body size (Fig. 5a, effect food N : P ratio on Daphnia body size: df = 13, p < 0.001). Throughout the experiment, food N : P ratio had a significant effect on the association intensities of A. parasiticum as well as of all epibionts pooled (Table 5). In both cases, association intensities decreased with increasing food N : P ratios, especially at the highest N : P level (Fig. 5c,d). Furthermore, food N : P ratio affected the prevalences of A. parasiticum and Vorticella sp., but only on day 20 (Table 5). On that day, prevalence of A. parasiticum was highest at intermediate food N : P ratio, whereas for Vorticella highest prevalence was found at the highest level of food N : P ratio.

Discussion
We investigated whether nutrient availability correlated with epibiont association intensity, prevalence, and richness in a field survey of natural D. pulex populations. We used phytoplankton N : P ratio (P-limitation) and N-concentration as proxies for food quality and found that food quality correlated with epibiont prevalence and infracommunity richness. An increasing phytoplankton N : P ratio (P-limitation) and N-concentration significantly reduced epibiont prevalence and infracommunity richness, i.e., the number of epibiont species infecting one zooplankton individual. The results of our field survey showed that averaged over all studied species, epibionts perform best in terms of prevalence and richness when P-availability in the environment is high and N-availability is low. The strong relationship with both nutrient availabilities shows that both an excess of N-and P-limitation contribute independently to the epibiont variables. Food quantity did not have a strong effect on the epibiont variables.
We next performed an additional laboratory experiment in which the obtained field results on epibiont variables were largely confirmed. Under P-limitation, food quality for the epibionts was less good, the Daphnia performed less well and epibiont association intensities were lower. For epibiont prevalence, the laboratory experiment yielded less clear evidence: significance in the results was limited to specific dates, and there were epibiont species specific effects. Prevalence of Vorticella increased under P-limitation. For A. parasiticum, the pattern was different with the highest prevalence being at intermediate N : P levels.
It is difficult to disentangle whether the detected nutrient change effect on epibiont variables is a direct effect of the algae on the epibionts, or an indirect effect of the algal nutrient changes on the epibiont community mediated by the Daphnia host. Interactions with the host may have played a role through different mechanisms. Host individual performance (Narr and Krist 2014;Reyserhove et al. 2017a), population density  and community composition changes (Johnson et al. 2007;Aalto et al. 2015) can all impact host-epibiont interactions. First, changes in Daphnia performance mediated by stoichiometric constraints (Wagner et al. 2017;Reyserhove et al. 2017a,b) may have played a role in addition to direct effects of stoichiometry on the epibionts. This is very likely, given that some of the epibionts (e.g., A. parasiticum and Colacium sp.) do not directly feed on the algae. In addition, competitive interactions for the algae between Daphnia and epibionts such as Vorticella and Brachionus (e.g., less P-availability upon increased N) may have influenced the pattern (Aalto et al. 2014;Pauwels et al. 2014). Third, host population density and associated host-epibiont contact rates may have been altered, changing association dynamics (Aalto et al. 2015). Increasing host population density positively impacted epibiont prevalence and infracommunity richness in the field study. It can be expected that, when host population density increases, epibiont prevalence, transmission, and infracommunity richness increases as well (Kiffner et al. 2014;Sponchiado et al. 2016). Thus, the D. pulex dynamics may have driven the epibiont dynamics in the field. However, no significant correlation between algal nutrient content or N : P ratio and D. pulex population density could be detected in our data. Also, microbial symbionts have been described to play an important role in food acquisition (Sison-Mangus et al. 2015;Callens et al. 2016) and toxin degradation of cyanobacterial algae (Macke et al. 2017) in Daphnia. Especially with respect to nutrient acquisition in zooplankton, microbial symbionts and epibiotic microflora may play a structuring role as catalysts for biogeochemical reactions involved in nutrient recycling and eutrophication (Eckert and Pernthaler 2014;De Corte et al. 2018;van de Waal et al. 2018).
In conclusion, we here found an effect of a food quality gradient on epibiont prevalence and association intensity in a field survey, which was partly confirmed in a laboratory experiment in which the nutrient ration in the algae was manipulated. Overall, a low phytoplankton N-concentration and N : P ratio/high P-availability was associated with higher epibiont prevalence, co-associations and association intensity. The discrepancy between laboratory results and results of the field survey for prevalence might indicate that the pattern for prevalence in the field is in part dictated by differences in Daphnia population density. Table 5. Statistical analysis on the impact of food N : P ratio on epibiont variables (prevalence and average association intensity of A. parasiticum, Vorticella sp., and both epibionts combined) in the Daphnia cohort experiment. Significant data (p < 0.05) is indicated in bold. Highly significant data (p < 0.001) is italics.