Introduction
Human activities have substantially increased the flux of nutrients from land into receiving waters (Smith, 2003). This nutrient enrichment enhances aquatic primary production, and may lead to dramatic changes in the composition and structure of aquatic food webs (Schindler and Fee, 1974; Smith et al., 2006). Specifically, an increased nutrient supply might promote development of harmful cyanobacterial blooms (Paerl et al., 2001, 2011; Smith and Schindler, 2009). Although nutrient loading has increased, primary production in aquatic ecosystems is often still limited by nitrogen (N) and/or phosphorus (P) (Elser et al., 2007; Bracken et al., 2014). This may be a result of an imbalanced nutrient supply (Carpenter et al., 1996; Sterner et al., 2007), as well as an increased nutrient demand associated to high phytoplankton densities (Carpenter et al., 1996). Nutrient limitation will alter the elemental composition of phytoplankton, and may specifically increase carbon:nutrient ratios (Sterner and Elser, 2002). As a consequence, nutritional quality of the phytoplankton decreases, thereby possibly constraining higher trophic levels (Sterner and Elser, 2002; Hessen et al., 2013). This may particularly apply to parasites that solely depend on their host as a food source (Smith, 2007).
Fungal parasites are very common pathogens infecting phytoplankton (Gerphagnon et al., 2015), which represent an important but yet overlooked ecological driving force in aquatic food web dynamics (Sime-Ngando, 2012). These parasites, belonging to the phylum Chytridiomycota and often referred to as chytrids, are host specific zoosporic fungi that can parasitize on phytoplankton and completely rely on their host to obtain energy and nutrients leading to death of the host (Sparrow, 1960, 1968; Barr, 2001). Thereby, they play an important role in natural aquatic ecosystems, in which chytrids can significantly change phytoplankton abundance and seasonal succession (Reynolds, 1973; Van Donk, 1984, 1989). Additionally, the free swimming stage of the chytrids (i.e., zoospores) may provide higher trophic levels with an alternative food source during blooms of large inedible diatoms (Kagami et al., 2007; Frenken et al., 2016) or cyanobacteria (Agha et al., 2016). Earlier work indicates that zoospores might find their host by chemotaxis (Muehlstein et al., 1988), and penetrate host cells using a rhizoidal system through which nourishment is conveyed to the zoospore (Van Donk and Ringelberg, 1983). After infection, the spore forms a sessile stage (i.e., sporangium) in which new zoospores (up to 60) are produced (Canter and Lund, 1951; Sparrow, 1960).
Chytrid zoospores generally contain a relatively high amount of nucleic acids that are particularly rich in P, but also contain substantial amounts of lipids, including fatty acids and sterols, which are rich in carbon (Barr and Hadland-Hartmann, 1978; Beakes et al., 1988, 1993; Elser et al., 1996; Kagami et al., 2007). Chytrids thus seem to have high P demands, as has been indicated by their low C:P as compared to their host (Kagami et al., 2007). As a consequence, limitation by P may affect a chytrid more than its host. Chytrid infections were indeed shown to be affected by host P limitation. More specifically, chytrid growth rate and the number of zoospores per sporangium decreased, and, as a function of lower host growth rate, zoospore loss as well as searching time increased, as compared to non-limited conditions (Bruning, 1991). If, however, P limitation impedes algal growth to a greater extent than that of the chytrid, epidemics may still occur (Bruning and Ringelberg, 1987; Bruning, 1991).
Nutrient limitation not only alters growth and reproduction of the parasite, it may also affect host defense. Freshwater cyanobacteria produce a wide range of oligopeptides including toxic microcystins (MC) (Welker and Von Döhren, 2006), which have been associated to chytrid defense (Rohrlack et al., 2013). These oligopeptides are N rich compounds, and their synthesis is typically constrained under N limitation (Van de Waal et al., 2010, 2014). Thus, during low N:P conditions, host defense may be reduced and thereby facilitate chytrid infections. In contrast, cellular N may accumulate under high N:P conditions and thereby enhance host defense. Limitation by N and P may thus have contrasting effects on chytrid infections of cyanobacteria. We hypothesized that parasite elemental composition will follow changes in the stoichiometry of its host, and that its reproductive success will decrease with host nutrient limitation. To test this hypothesis, we exposed the cyanobacterium Planktothrix rubescens to its chytrid Rhizophydium megarrhizum under control, low N and low P conditions, leading to a range of host N:P ratios. We predict that infections will decrease with increasing host N:P, as the availability of P for chytrid nutrition will decrease and the host defense by oligopeptides will increase.
Discussion
Although the different nutrient supply ratios had only minor effects on Planktothrix growth rates after acclimatization (Table 1), there were clear changes in the elemental composition of the cyanobacteria (Figure 2A). This indicates that nutrient depletion did affect host physiology, but not growth, at the time of sampling. Apparently, Planktothrix is able to maintain similar maximum growth rates as compared to the control at both a high and low N:P supply ratio (Table 1). The reduced host N:P under low N:P supply and increased host N:P under high N:P supply (Figure 2A) indicates nutrient limitation at the end of the exponential phase and/or early stationary phase. Moreover, population densities in the low N:P treatment at the end of the experiment were lower as compared to the control and high N:P treatment (Figure 1B). These lower population densities are mainly caused by a low N availability in the low N:P treatment, but may also result from the lower Planktothrix population densities at the start of the experiment. Differences in host population density may affect light availability in the cultures. A lowered population density, as observed in the low N:P treatment, may have resulted in an increased light availability due to reduced self-shading. Earlier studies have indicated that zoospores may find their host using chemical cues that are related to photosynthetic activity, since zoospores are generally attracted to carbohydrates, polysaccharides, proteins and amino acids (Muehlstein et al., 1988; Donaldson and Deacon, 1993; Moss et al., 2008). Some studies, however, also reported attachment of zoospores to new hosts during dark conditions (Barr and Hickman, 1967). Indirectly, the relative higher light availability in the low N:P treatment may thus have favored parasite attraction. Moreover, with a comparable amount of zoospores added at the start of the experiment, the zoospore-to-host ratio was also higher in the N limited treatment, which may favor infection rates. Yet, both the zoospore production rate and production efficiency were lower in the low N:P treatment, and did not lead to a different infection rate as compared to the control (Table 1). This suggests that higher relative light availabilities as well as higher initial zoospore-to-host ratios did not stimulate, and possibly even impeded the infection dynamics in our low N:P treatment.
In response to an increased N:P supply in the medium, host N:P increased as well. This resulted in a consecutive increase in the N:P of the zoospores (Figure 2B). Our results thus show that stoichiometry of a host can cascade to their parasites. In the host, changes in stoichiometry seem to be driven mainly by a change in P content, as host C and N remain constant with changing N:P supply (Figure 2C). In the parasite, changes in zoospore stoichiometry are also mainly driven by P content, but C and N contents decrease as well. Yet, P content decreases faster, suggesting a higher flexibility of the chytrid with respect to P (Figure 2D). These findings indicate that chytrid parasites can be stoichiometrically flexible, while maintaining their ability to infect along an N:P supply gradient. N:P and C:P ratios of the chytrid used in this experiment are relatively high as compared to two other studies using different chytrid species (Kagami et al., 2007, 2017), but fall well within the range of aquatic fungi reported before (Danger and Chauvet, 2013; Danger et al., 2015). These data furthermore suggest that fungal elemental homeostasis is indeed limited (Persson et al., 2010; Danger et al., 2015). Zoospore N:P ratios largely resembled that of the host under control conditions, but were different from the host in the low and high N:P treatment. We could not separate zoospores from the heterotrophic bacteria, and our results may therefore have been confounded by shifts in bacterial numbers. To prevent high bacterial numbers fueled by the lyses of Planktothrix, we ran the experiments over a relatively short time period. Consequently, the overall biomass of bacteria and thereby their contribution to the elemental composition at the time of sampling is likely to be small.
Increasing N:P supply ratios resulted in an increased zoospore production rate and production efficiency (Figures 4A,B). These results are in contrast to earlier findings of an experiment that showed that under P-limitation (and presumably high N:P) zoospore production decreased (Bruning, 1991). It might be possible that the chytrid species (R. planktonicum Canter emend.) used in the experiment by Bruning (1991) has higher P-requirements, and therefore suffered more from P-limitation. Or, because the chytrid in our experiment can potentially infect and exploit multiple adjacent cells within one cyanobacterial filament (Canter and Lund, 1951), it might be less vulnerable to nutrient limitation. In other words, the chytrid might continue infecting adjacent cells until it has consumed sufficient nutrients to complete an infection cycle. However, this is only profitable if the energetic costs of growing rhizoids and producing degrading enzymes to invade host cells balance the gains with respect to resource acquisition.
Under low N:P conditions, the parasite zoospores were fewer but contained more C, as well as N and P as compared to the control (Figure 2D). Although we did not assess zoospore size in this experiment directly, increases in elemental contents do suggest that the chytrid produced larger zoospores. Zoospore size of the used chytrid was shown to vary from 2.84 to 5.36 μm under control growth conditions (Supplementary Material), and variation in spore size was also shown in other studies describing shifts in spore size with climatic conditions (Kauserud et al., 2008, 2011). Presumably, larger zoospores facilitate zoospore survival time, since they can contain more lipids and fatty acids that might represent an energy source to fuel zoospore metabolism (Steinhoff et al., 2011). A longer spore survival time may be particularly favorable at lower host densities, and may explain the unaltered chytrid infections in the low N:P treatment. Conversely, in the high N:P conditions, more zoospores were produced per host biomass but contained less C, N and P per zoospore, suggesting that they were smaller. These findings are supported by earlier observations indicating that the efficiency of spore production by a parasitic dinoflagellate is increased under high N:P conditions, which might result in a higher transmission to new hosts under high host density conditions (Yih and Coats, 2000).
At low N:P supply the chytrid seems to produce a low amount of large zoospores, while at high N:P supply it produced a higher amount of small zoospores. The chytrid thus possibly produced smaller spores with a higher production efficiency (Figure 4C), suggesting a trade-off between size and production rate as well as success of infection. In other words, larger spores may survive longer providing the chytrid more time to find a suitable host under low host density conditions, while smaller ones survive less long but due to their high numbers achieve a higher infection transmission in high host density conditions. A trade-off between organism size and growth rate has also been reported for various other organisms, including phytoplankton (Nielsen, 2006) and zooplankton (Stemberger and Gilbert, 1985). Moreover, a trade-off between zoospore survival time and production rate was observed in another chytrid, the amphibian killing fungus Batrachochytrium dendrobatidis (Woodhams et al., 2008). Thus, changes in host N:P stoichiometry may affect the growth strategy of the parasite, following a more general trade-off between cell size and production rates (Figure 4C). Such changes can have consequences not only for the infection dynamics, but also for higher trophic levels that are provided with either many smaller zoospores, or fewer larger ones.
As expected, intracellular MC content closely followed the relative availability of N, and thus increased with cellular N:P ratios (Figure 5). These results are in line with earlier work, showing a strong dependency of MC contents on N availability (Van de Waal et al., 2009, 2010). In the treatment with high N:P supply and high MC production, however, zoospore production (Figure 4A) and infection rate (Table 1) were highest. Additionally, in the low N:P treatment, MC contents was lowest while zoospore production and infection rate were not different from the control. So, there was no clear relation between intracellular MC content and chytrid proliferation. Possibly, a considerable fraction of the total MC might be bound to proteins of cyanobacterial cells (Zilliges et al., 2011), which were not included in our extraction processes. Furthermore, Planktothrix may produce other oligopeptides that play a role in parasite defense systems (Sønstebø and Rohrlack, 2011; Rohrlack et al., 2013), which were not analyzed here.
The intracellular MC seemed to be lower in the parasite exposed treatments as compared to the unexposed treatments. This may possibly result from leakage of MC from the cells into the liquid phase (Jones and Orr, 1994). Moreover, chytrid rhizoids that invade the host cells might use enzymes that are able to digest MC. Indeed, fungi were shown to be capable of degrading MC (Jia et al., 2012). In our experiment, however, extracellular MC concentrations nor chytrid MC contents were analyzed. If MC is released into the water column from cyanobacterial cells, it can have consequences for other organisms present (Carmichael, 1992; Zurawell et al., 2005). For instance, high MC concentrations in the water can accumulate in Daphnia (Chen et al., 2005) and have adverse effects on growth and development of fish (Jacquet et al., 2004). Yet, actual exposure of other organisms to MC in the water may be limited, as MCs can be rapidly biodegraded and detoxified by bacteria and adsorb to plants and sediments (Harada and Tsuji, 1998; Pflugmacher et al., 2001; Kato et al., 2007). Whether MCs can bind to- or be transported into zoospores is unknown. But, if this would occur, zooplankton might be exposed to MCs via this indirect route, since zoospores can serve as a food source for copepods, cladocerans and possibly rotifers (Kagami et al., 2004, 2007, 2011; Buck et al., 2011; Agha et al., 2016; Frenken et al., 2016).
Our results demonstrate an increase in infection rate with host N:P stoichiometry, thereby showing the opposite to what we hypothesized. Because chytrids seemed relatively more P rich as compared to their host, we initially predicted that host P content would constrain chytrid growth more than it would constrain the host (Bruning and Ringelberg, 1987; Bruning, 1991). Our results suggest, however, that chytrid proliferation is much more sensitive to the relative availability of N. Specifically, if this increases (i.e., higher host N:P), infection rates increase, while if this decreases (i.e., lower host N:P), infection rates decrease. This is also shown by the lower flexibility of the parasite N content as compared to P, suggesting that spores are more likely to be constrained under low N conditions. It remains unclear why infection rates increase under P limitation and relative high N contents. Particularly as under these high N:P conditions, MC contents were highest as well. We initially expected that under such conditions, chytrid infections can be inhibited by MCs in its host. The increase in MC content with high N:P conditions, however, was relatively small and may therefore not have been sufficient to inhibit the chytrid infection. Possibly, regulation of other oligopeptides in response to N:P supply could have explained the observed responses, and should thus be included in future analyzes. Moreover, other metabolites synthesized by cyanobacteria under high N:P supply may have facilitated chytrid growth and reproduction. Further detailed biochemical analyses of chytrids and their distinct developmental stages would be required to fully understand the stoichiometric interactions with their hosts, and particularly the putative important role of N in controlling infections.
Our analysis revealed some still poorly understood effects of nutrient availability on the interaction of a host–parasite system. Shifts in nutrient supply ratios not only lead to a shift in host stoichiometry, but also to comparable changes in the parasite. Thereby, we show that elemental stoichiometry of a host can cascade to their parasites. We hypothesize that, in response to changes in nutrient supply, the parasite may exhibit a trade-off between size and zoospore production rate to optimize reproductive success. Therefore, nutrient limitation may indirectly affect parasite abundance and stoichiometry. Since chytrids can facilitate growth of zooplankton (Kagami et al., 2007; Agha et al., 2016), changes in parasite production, stoichiometry and cell size may have implications for aquatic food web dynamics.