Daphnia performance on diets containing different combinations of high‐quality algae, heterotrophic bacteria, and allochthonous particulate organic matter

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2020 The Authors. Freshwater Biology published by John Wiley & Sons Ltd 1Department of Ecology and Environmental Science, Umeå University, Umeå, Sweden 2GEOMAR Helmholtz Centre for Ocean Research Kiel, Kiel, Germany 3Department of Aquatic Sciences and Assessment, Swedish University of Agricultural Sciences, Uppsala, Sweden


| INTRODUC TI ON
Zooplankton production in lakes is supported by both autochthonous and allochthonous carbon sources. While the classic concept of the pelagic food web regards phytoplankton (autochthonous) production as the major carbon source (e.g. Nauwerck, 1963;Wetzel, 2001), there are several studies that suggest that terrestrial (allochthonous) carbon is an important subsidy for higher trophic levels (e.g. zooplankton and fish) in unproductive lakes (e.g. Cole et al., 2011;Pace et al., 2004). However, the importance of this allochthonous subsidy has been questioned by studies that show that allochthonous carbon is inefficiently incorporated in the food web (Brett et al., 2017;Lau et al., 2014).
Terrestrial carbon enters lakes in the form of dissolved (t-DOM) or particulate organic matter (t-POM) (Cole et al., 2006). Both DOM and POM can enter the food web, either by uptake of t-DOM by heterotrophic bacteria Tranvik, 1988) which are then grazed by other heterotrophic organisms such as flagellates, ciliates, and metazooplankton Jones, 1992), or by direct feeding on t-POM (Cole et al., 2006;Hessen, 1998;Pace et al., 2004). Terrestrial OM is often dominated by recalcitrant dissolved and particulate OM that has been degraded in the catchment before being exported to lakes (Wetzel, 1984). The chemical character of the t-POM varies depending on the characteristics of the terrestrial ecosystem but most of this OM is expected to have a low nutritional value for metazoans due to its high carbon:phosphorus (C:P) stoichiometry and lack of essential biochemicals. This has been shown for different materials, for example grasses and deciduous tree litter (Brett et al., 2009;Taipale et al., 2014Taipale et al., , 2016 and t-POM from boreal forest peat layers (Wenzel et al., 2012a).
The magnitude of t-OM inputs into boreal lakes depends both on catchment characteristics and season. Inputs of t-OM are usually high during spring flood and high-flow episodes (Bergström & Jansson, 2000;Hope et al., 1994;Jonsson et al., 2001) and the proportion of t-POM increases with increasing water discharge (Thurman, 1985). Generally, t-POM accounts for <10% of t-OM (Ivarsson & Jansson, 1994;Thurman, 1985). However, as the total input of t-OM can be very high in many boreal lakes, the particulate fraction nevertheless constitutes a large amount compared to other potential food sources. Concentrations of dissolved organic carbon in humic lakes can range between 20 and 50 mgC/L (Thurman, 1985).
Hence, particulate organic carbon in those lakes may be as high as 5 mgC/L of which phytoplankton and bacteria only account for a small portion . In contrast, boreal lakes with high inputs of t-OM often only support a low phytoplankton biomass and algal carbon concentrations rarely exceed 0.1 mgC/L Jansson et al., 2003).
The high concentrations of coloured, terrestrial material negatively affect the light climate in humic lakes, resulting in low primary production and biomass of phytoplankton (Deininger et al., 2017;Seekell et al., 2015). As t-OM can support high bacterial production (Hessen, 1985;Jansson et al., 2003), boreal humic lakes are often characterised by high bacteria:phytoplankton biomass ratios compared to other systems (Hessen, 1998) with bacterial biomasses being up to 3-times higher than phytoplankton biomass (Hessen, 1992;Hessen & Andersen, 1990;Jansson et al., 2003). Heterotrophic bacteria are directly consumed by a variety of non-selective filter feeding zooplankton, e.g. Daphnia (Wylie & Currie, 1991). Owing to their low C:P stoichiometry, bacteria may be considered a high-quality food for organisms with high somatic P content, e.g. daphnids (Sterner & Elser, 2002). However, bacteria do not meet consumers' requirements in terms of biochemical food quality because they lack essential polyunsaturated fatty acids (PUFAs) (Zelles, 1999) and sterols (Martin-Creuzburg et al., 2011;Volkman, 2003). Consequently, pure bacterial diets cannot sustain Daphnia populations (Taipale et al., 2012;Wenzel et al., 2012b).
As phytoplankton, heterotrophic bacteria, and t-POM are present in varying proportions in lakes, zooplankton will encounter fluctuating food qualities and quantities throughout their lives. Therefore, it is necessary to investigate how food quality and quantity interact in the regulation of secondary productivity. Some studies suggest that food quality does not play an important role at low food concentrations when energy limitation dominates (Elser et al., 2003;Persson et al., 2007;Sterner & Robinson, 1994). However, other studies report that food stoichiometry does affect grazer performance irrespective of food quantity (Boersma & Kreutzer, 2002). However, these studies have mainly focused on algal food resources of varying quality, so where the alternative food resource is non-algal it remains unclear how food quality and quantity interact. Wenzel et al. (2012a) concluded that a substantial fraction of algal carbon was needed for Daphnia growth and reproduction when the food consisted of algae supplemented with bacteria or t-POM. Although the study clearly showed that bacteria and in particular t-POM are low-quality food resources that cannot sustain Daphnia populations alone, the study did not test to the extent to which the quantity of the high-quality algal food can offset the negative impact of the low-quality food.
The main objectives of this study were: (1) to investigate how basic life history parameters of a non-selective filter feeder (Daphnia 6. Therefore, we conclude that heterotrophic bacteria, but not peat layer t-POM, can be an important component of zooplankton diets in boreal lakes, especially if the concentration of phytoplankton is low.

K E Y W O R D S
boreal forest detritus, Daphnia galeata, food quality, Pseudomonas sp., Rhodomonas lacustris galeata) are affected if high-quality food (phytoplankton) is diluted with different types of low-quality food (bacteria or t-POM, respectively); and (2) to determine to the extent to which the observed response depends on the absolute quantity and proportion of the high-quality food. In contrast to the majority of food quality studies, we used both non-limiting as well as limiting concentrations of both high-and low-quality food resources. Based on previous results (Wenzel et al., 2012a(Wenzel et al., , 2012b, we hypothesised that increasing proportions of bacteria and t-POM in the diet will lead to decreased survival, somatic growth, and reproduction of Daphnia despite the presence of phytoplankton and that these effects will be more pronounced for t-POM than for heterotrophic bacteria. We also hypothesised that this response will be stronger when the concentration of phytoplankton is limiting.

| Experimental design
We conducted experiments with Daphnia feeding on different concentrations of high-quality phytoplankton that were mixed with either heterotrophic bacteria or t-POM. In each experimental run, the food suspension contained a constant Rhodomonas concentration and a range of concentrations of either Pseudomonas or t-POM. The different Rhodomonas concentrations were low (0.22 mgC/L), intermediate (0.37 mgC/L), and high (0.55 mgC/L), which is either below (low), at (intermediate), or above (high) the incipient limiting level above which ingestion rate is maximal (0.26-0.36 mgC/L for similar sized D. longispina, Lampert, 1987). To each of the three Rhodomonas concentrations, we added five different concentrations of either Pseudomonas or t-POM (target concentrations 0.15, 0.3, 0.6, 1.2, and 2.4 mgC/L). In each experiment we also ran a treatment with Rhodomonas only. This resulted in a total of 33 treatments (three Rhodomonas concentrations × two additional food types × five addition levels, plus threee treatments with Rhodomonas only). Due to space constraints we performed three experimental runs in sequence, one for each Rhodomonas concentration.
A clone of D. galeata isolated from a boreal mesohumic lake, Nydalasjön (63°49'N, 20°20′E), was grown in 3 × L16 medium (Lindström, 1991) and fed chemostat-grown (see below) Rhodomonas ad lib. prior to the experiments. Care was also taken to avoid extensive crowding of Daphnia in the cultures. For the experiments, we used neonates from the third clutch or later that had hatched within 24 hr. For each treatment, we used between seven and 11 replicate daphnids that were kept individually in approximately 25 ml feeding suspension in glass scintillation vials (total n = 384 individuals). We transferred experimental animals into clean vials with fresh feeding suspension daily using a wide-mouthed pipette. Experimental Daphnia were reared under the same conditions, kept in the temperate-controlled rooms with identical conditions during all experimental runs. Survival, occurrence of eggs, and number of neonates hatched were recorded on these occasions. Neonates were then removed. Each experiment was run for 10 days. Even though care was taken to avoid air bubbles when closing the vials, occasionally bubbles developed overnight and some daphnids died after being trapped in the surface tension. These individuals were randomly distributed across treatments. Because it cannot be assumed that feeding and reproduction is unaffected by being caught in the surface tension, we excluded these individuals from further analysis (n = 32 individuals).
Individual daphnids were photographed in a dissecting microscope at day 5 and the body length was measured from the top of the helmet to the base of the caudal spine. Dry mass was then calculated using a pre-determined relationship between dry mass (DM, µg) and body length (L, µm) for this clone (lnDM = 3.96lnL -25.4).
Juvenile specific growth rate, r, was calculated as: where α 0 is the average initial dry mass, M t is the dry mass of surviving individuals in each treatment, and t is the experimental duration (5 days) for this measurement.

| Rhodomonas and Pseudomonas cultures
We used the cryptophyte Rhodomonas lacustris (strain NIVA 8-82; from the Norwegian Culture Collection of Algae, maintained by the Norwegian Institute for Water Research and the University of Oslo, Oslo, Norway). The genus Rhodomonas has been revised and the species R. lacustris has been renamed Plagioselmis lacustris (Pascher and Ruttner) Javornicky (Javornický, 2001). However, we have chosen to keep the name Rhodomonas lacustris here because it has not been renamed in the culture collection. The Rhodomonas was cultured in 4-L chemostats in modified 3 × L16 medium adjusted to pH c. 8, containing 100 µl/L Wright's solution, and vitamins and animal trace elements as in COMBO medium (Kilham et al., 1998). We chose Rhodomonas as it is a high-quality food for zooplankton, supporting high growth and reproduction rates (Brett & Müller-Navarra, 1997), and because cryptophytes often constitute a large fraction of the phytoplankton biomass in humic lakes (Deininger et al., 2017;Sarvala et al., 1999). Chemostats were run in a 16:8-hr light:dark cycle at an ( 10 µm in diameter, which is comparable to the size of Rhodomonas cells. The t-POM suspensions were prepared fresh every second or third day by determining carbon content (as described above) and

| Preparation of t-POM
diluting with 3 × L16 medium to the desired concentrations.

| Chemical analyses
Samples for particulate P analysis of Rhodomonas, Pseudomonas, and t-POM suspensions were collected daily (per batch for t-POM) during the experiments on acid-rinsed GF/F glass fibre filters (25 mm, Whatman, Whatman Arbor Technologies, Ann Arbor, MI, U.S.A.) and analysed with the molybdate-blue method after hot acid digestion with potassium persulfate (Menzel & Corwin, 1965). Samples for particulate C and N analysis were collected at the same time on pre-combusted GF/F filters and analysed on a Costech ECS 4010 elemental analyser (Costech International S.p.A., Milano, Italy). To model juvenile specific growth rate, r, we used three different models, accounting for the effects of food quantity and food quality alone, and the interaction between food quantity and quality:

| Statistical analyses
where a, b, and c are the parameters of the models, totC is the total food concentration (mgC/L), %Rho is the % carbon contribution of Rhodomonas in the diet, which is used as a proxy for food quality.

| Survival
Daphnia survival over the 10-day experiment ranged between 0 and 100% across all treatments and was generally higher in treatments with Pseudomonas than t-POM when comparing at the same %Rhodomonas or at the same food quantity (Figure 1a

| Juvenile specific growth rate
The specific growth rate increased from 0.25 to 0.49 day -1 with increasing Rhodomonas concentration in the treatments with Rhodomonas only, but showed a more complex response to food quantity when either Pseudomonas or t-POM was added, with apparent effects of both food quantity and quality ( Figure 2). In the high Rhodomonas treatment, addition of Pseudomonas initially did not affect the growth rate but at larger additions of bacteria, the growth rate declined (Figure 2a  negative effect of the addition of bacteria on the growth rate, but the decline began at small additions of bacteria (Figure 2a,c). In contrast, there was a unimodal response to the addition of bacteria in the low Rhodomonas treatment, with the highest growth rates observed at the smaller additions of bacteria (Figure 2a,c). In the t-POM treatments, the growth rate generally decreased with increasing amounts of t-POM added (Figure 2b,d).
The model that best described growth rate response in the experiment was the one including the interaction between food quality and quantity (eq. 4; Table 3; Figure 2; Appendix S1). This model was highly significant and explained 76 and 64% of the variation in the Pseudomonas and t-POM treatments, respectively. The models containing only food quality or food quantity (Eq. 2, 3) were mostly non-significant and explained little or nothing of the variation in growth rate (Table 3; Appendix S1). The decline in growth rate with increasing food quantity and decreasing food quality was steeper when adding t-POM than Pseudomonas (Figure 2), which is also reflected by a larger value of the parameter c in Eq. 4 for t-POM than for Pseudomonas (Table 3).

| Reproduction
The total production of eggs and juveniles during the experiment was positively affected by the interaction between %Rhodomonas in the diet and total food concentration. Both the main effects,

| D ISCUSS I ON
Our results suggest that the availability of high-quality phytoplankton biomass is critical for Daphnia survival, growth, and reproductive output. As hypothesised, an increasing dilution of a non-limiting concentration of high-quality Rhodomonas with low-quality Pseudomonas or t-POM led to decreased overall Daphnia performance. We also found strong interacting effects between food quality and quantity:  (Taipale et al., 2012;Wenzel et al., 2012b). Likewise, at high Rhodomonas concentration, the addition of bacteria could initially sustain a high growth rate, which eventually declined at large additions of bacteria. It has previously been shown that Daphnia can readily ingest and assimilate bacterial C and P (Hessen & Andersen, 1990;Hessen et al., 1989;Taipale et al., 2012;Wenzel et al., 2012b) and that dietary P is especially important for high-P requiring organisms such as Daphnia (Sterner & Elser, 2002). However, while providing enough energy (organic C) and minerals (P), heterotrophic bacteria in general lack essential fatty acids (FAs) (Zelles, 1999) and sterols (Martin-Creuzburg et al., 2011;Volkman, 2003), which is probably terial and concluded that t-POM only played a minor role for zooplankton nutrition while phytoplankton was a much higher quality resource. However, as daphnids did survive and even produced viable eggs on leaf-dominated diets, the authors also concluded that the quality of their t-POM was higher than that of prokaryotic cyanobacteria, which are usually considered a low-quality food (von Elert et al., 2003;Martin-Creuzburg et al., 2008). In contrast to Brett et al. (2009), our peat layer t-POM supported lower Daphnia survival, growth, and reproduction than the prokaryote Pseudomonas in a previous study, especially at an overall low food concentration (Wenzel et al., 2012a). Likewise, the present study shows that somatic growth responses differed substantially between Pseudomonas and t-POM additions in the low Rhodomonas treatments. While t-POM additions caused decreased growth, supplementing Rhodomonas with Pseudomonas initially resulted in an increase in growth rates, followed by a gradual decline in growth rate with increasing addition of Pseudomonas. Compounds that are toxic to daphnids (e.g. phenols; Taylor et al., 1996) may also have been present in our t-POM. Since we did not measure the concentrations of any potentially toxic compounds, we cannot know to what extent it may have contributed to the strong negative response of the daphnids to increasing t-POM concentrations. Nevertheless, the t-POM we used was intentionally selected to mimic a type of t-POM that occurs naturally in boreal lakes.
Regardless of the exact mechanism behind the strong negative effect of t-POM, it is safe to conclude that t-POM of this type does not sustain survival, growth, or reproduction. We know from previous experiments that a 100% pure diet of our peat layer t-POM, or Pseudomonas, cannot sustain survival, growth or reproduction of D. galeata (Wenzel et al., 2012a, b). Thus, one could therefore argue from our results in the Pseudomonas treatments (i.e. similar to the arguments by Brett et al. (2009) for t-POM;and Taipale et al. (2014) for heterotrophic bacteria and t-POM), that even low levels of phytoplankton in Daphnia diets may greatly improve their ability to use low-quality resources.
However, for bacteria, it has to be noted that the quality of bacterial strains and taxa may differ substantially. Taipale et al. (2012) compared the nutritional value of three bacterial taxa and reported different growth and reproduction responses of Daphnia. A study by Freese and Martin-Creuzburg (2013) showed that some bacterial strains may even be toxic for zooplankton, notably a strain of Pseudomonas. The presence of hormesis, where opposite effects in small and large doses of an exposure are seen in organisms due to compensation for an initial disruption/damage caused by a toxin (Stanley et al., 2013), might also be the reason for the beneficial effect in Daphnia growth noted when low concentrations of bacteria were added to the low phytoplankton treatment (cf. Figures 1-2).
However, we cannot conclude from our experiment whether indigestibility, poor content of essential nutrients and/or biochemicals, or toxicity reduces food quality for our D. galeata clone, so this remains to be tested.
The ability to use bacterial carbon in the presence of phytoplankton may also decrease over time. For example, the assimilation of cyanobacterial carbon by Daphnia has been shown to be dependent on essential lipid reserves, leading to a decline in carbon assimilation over time when feeding on cyanobacteria-dominated diets (DeMott & Muller-Navarra, 1997). This mechanism may as well apply to diets dominated by heterotrophic bacteria and explain why mortality in our study increased during the second half of the experiment when low amounts of Rhodomonas were supplemented with Pseudomonas.
Thus, bacteria can temporally be an important subsidy to overcome periods of low phytoplankton abundances. However, in the long run, this will not be enough to support population growth.
Somatic growth rates alone are not enough to evaluate food quality as organisms often have different requirements during different life stages (Sterner & Schulz, 1998). In Daphnia, juveniles show a higher sensitivity to mineral deficiencies (P) (Urabe & Sterner, 2001), while biochemical limitation plays a more pronounced role for reproductive output in later life stages (Becker & Boersma, 2003). In agreement with these findings, juvenile survival was lower with the addition of t-POM with low P content compared to P-rich Pseudomonas. Differences in egg and juvenile production, on the other hand, were less pronounced between t-POM and Pseudomonas addition treatments which both are deficient in FAs (Wenzel et al., 2012a) and probably also sterols. This underlines the importance of biochemicals for reproductive success.
Polyunsaturated FAs especially may be important for egg production in daphnids, and the minimum PUFA concentrations needed for successful reproductive output in natural systems may be even higher than previously anticipated (Ravet et al., 2012).
While our medium and high Rhodomonas experiments may be representatives of productive, nutrient-rich eutrophic lakes, the low Rhodomonas concentration tested resembles conditions that zooplankton can experience in humic lakes where phytoplankton biomass seldom exceeds 0.1 mgC/L . Here, when feeding on Rhodomonas only, daphnids were energy limited and did not perform as well as with moderate additions of Pseudomonas.
These intermediate additions in the low Rhodomonas treatments are similar to conditions in humic lakes where bacteria:phytoplankton biomass ratios of up to 3:1 have been reported (Hessen, 1992;Jansson et al., 2003). Thus, the presence of heterotrophic bacteria may not only increase food quantity but also quality for Daphnia by supplying essential minerals such as P as long as a certain concentration of high-quality phytoplankton is still available to meet biochemical needs. In contrast to the positive impact of supplementing Rhodomonas with Pseudomonas, the addition of t-POM negatively affected Daphnia performance in all treatments. Thus, high concentrations of t-POM seem to reduce the capability of Daphnia to use phytoplankton. This finding is relevant and has implications for zooplankton population growth in natural systems where t-POM can be present in high concentrations (up to 5 mgC/L in unproductive, humic lakes) (Thurman, 1985).
We used one species of phytoplankton in our study that is known to be a high-quality food for zooplankton (Brett & Müller-Navarra, 1997). In lakes, a mix of different species or taxonomic groups of phytoplankton will always be present and the composition of the algal community may undergo pronounced successional changes across seasons. As different algal taxa contain varying amounts of minerals and biochemicals and differ in their digestibility (Hessen & Leu, 2006;Sterner & Elser, 2002;Taipale et al., 2013), high-quality food may only be available in sufficient quantities during certain periods of the year. Therefore, zooplankton may in general benefit from mixed diets as different food components can constitute complementary resources, as illustrated in experiments feeding copepods (Twombly et al., 1998), cladocerans (Boersma & Vijverberg, 1995), and rotifers (Rothhaupt, 1991) with different phytoplankton. It can be expected that combinations of algal and non-algal food (e.g. heterotrophic bacteria) have similar effects, although it has not yet been thoroughly investigated.
We recognise that there can be considerable variation in the response to various environmental conditions among Daphnia species, among clones, and within clones (e.g. Gorokhova et al., 2002;Wolinska et al., 2007), but the response to food quality has been shown to be at least qualitatively similar among clones in a study using mixtures of low-and high-quality phytoplankton (Bednarska et al., 2014). Hence, we propose that our findings for this clone of D. galeata are likely to be qualitatively applicable to non-selective filter-feeding cladocerans in general, but not to selective suspension feeders such as diaptomid copepods. However, the quantitative response may not be the same across clones or species of cladocerans. This is also supported by stable isotope studies that report the use of more autochthonous food sources by copepods compared to cladocerans (Berggren et al., 2014;Karlsson et al., 2015). Hence, cladocerans will be probably more prone to food quality limitations than copepods when high-quality phytoplankton is heavily diluted in lake systems with high inputs of terrestrial material. It has also been shown that zooplankton in general show higher allochthonous stable isotope signals in systems with high abundances of heterotrophic bacteria (Karlsson et al., , 2007, which are either consumed directly (Karlsson et al., 2004) or via intermediate trophic levels such as ciliates and flagellates (Hiltunen et al., 2017;Jansson et al., 2007).
Hence, direct consumption of bacteria and/or via trophic upgrading via the microbial loop (Hiltunen et al., 2017;Jansson et al., 2007), and not direct t-POC consumption (this study, Taipale et al., 2014) seems to be the major pathway of terrestrial carbon to metazoan zooplankton.
In summary, our results show that the availability and amount of high-quality autochthonous phytoplankton production is a major factor that determines the performance of this clone of D.
galeata, and we suggest that qualitatively similar patterns may occur in other unselective filtering feeding cladocerans. We found that heterotrophic bacteria, but not peat layer t-POM, can be a temporally important component of Daphnia diets in boreal systems where the concentration of phytoplankton alone is limiting.
Terrestrial inputs of organic carbon may therefore support higher trophic levels by serving as a potential substrate for bacterial growth, either through direct grazing on heterotrophic bacteria (by cladocerans) or via intermediate bacterivorous consumers such as ciliates or flagellates (by cladocerans and copepods). In more nutrient-rich systems, however, where non-limiting concentrations of phytoplankton are present, both heterotrophic bacteria and t-POM might decrease especially cladocerans performance due to a dilution effect.

ACK N OWLED G EM ENTS
We thank Jan Johansson for the C/N analysis and Sebastian Diehl for asking an inspiring question after a seminar that led to the idea for this study. This study was supported by a grant from the Swedish Research Council to Mats Jansson and Tobias Vrede (2007-5523), and a grant from Tryggers foundation to Tobias Vrede.

CO N FLI C T O F I NTE R E S T
The authors have no conflict of interest to declare.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data are available from the authors upon reasonable request.