Ecology, Epidemiology, and Evolution of Parasitism in Daphnia
This chapter describes how to use the Daphnia parasite system for experiments. I first discuss the advantages of the system for research and education. Then I describe a number of experiments, some of which are very simple and are suitable for courses in experimental parasitology and ecology. The experiment section has two parts. The first discusses experiments in which the individual host is the unit of replication. Such experiments can be used to ask questions such as: How does a parasite affect its host? How is a parasite transmitted? The next part expands to discuss experiments that use entire populations as the unit of replication. Here I suggest experiments that pose questions such as: Does a parasite influence host density? Can a parasite drive its host population to extinction? How quickly can hosts evolve resistance?
The Daphnia–parasite system is particularly suitable for testing hypotheses because it allows for the creation of rather simple experiments. Among the advantages of this system are:
Under laboratory conditions (20°C), Daphnia produce their first eggs after 7-15 days (depending on the food level). This equals the shortest possible generation time in experiments. Thereafter, they produce a clutch of parthenogenetic eggs every 3-4 days until death, which results in an approximately constant fecundity across the adult life span. The first clutch is usually smaller than the following clutches. Only very low food levels may result in skipped clutches.
Controlled conditions allow other extrinsic sources of mortality, e.g., predation by fish, infection by other parasites, to be excluded.
Parthenogenetic reproduction allows the females to remain isolated (1 female in 30-200 ml of culture medium) so that fecundity and death schedules can be recorded accurately. From these, birth and death rates can be calculated in the absence of density dependence. Individual females can be kept with or without parasites.
Parthenogenetic reproduction further allows for the separation of genetic (among-clone variance components) and nongenetic effects (within-clone variance components).
Many Daphnia parasites fit the definition that epidemiological models use for microparasites very well: small, unicellular parasites that reproduce directly within their hosts and are directly transmitted among hosts (Anderson and May 1991).
Many parasites allow for the freezing of transmission stages, which means that the same genetic material can be stored for long time periods. Daphnia can be kept clonally for very long times (years!) and thus can be kept essentially without genetic changes for long periods.
For experiments with replicated populations (rather than individuals), the following points are also relevant:
The population growth of Daphnia in laboratory populations with a constant food supply and no parasites is reasonably well described by a logistic growth model. In the absence of parasites, Daphnia populations reach an equilibrium population level that represents carrying capacity. Daphnia populations have overlapping generations. Generation times in population experiments are about 10 to 15 days (at 20°C).
Clonal reproduction of hosts avoids complications attributable to mate choice and mate finding. It also excludes complications attributable to the effects of inbreeding or outbreeding.
Daphnia's planktonic way of life approximates well-mixed conditions without strong spatial structure. Transmission of most Daphnia parasites is through waterborne transmission stages. These follow the common epidemiological assumption of mass action nicely, which states that the likelihood of transmission is strictly a function of the population densities (or sizes) of infected and uninfected hosts.
Epidemiological models usually assume homogeneously mixed populations without genetic structure. Monoclonal Daphnia populations fulfill these criteria perfectly. This is advantageous for an experimental system because the absence of genetic host diversity allows one to exclude the confounding effect of host evolution, which may otherwise rapidly change the genetic structure of the populations (Capaul and Ebert 2003). In monoclonal Daphnia populations, genetic diversity can only arise by mutations, and mutation rates are too low to play a significant role in experimental Daphnia populations that are kept for a limited period (less than a few years).
Daphnia parasites usually produce persistent (chronic) infections. Unless the host clears the infection within the first 1 or 2 days after exposure, it will not recover from the infection. Therefore, only two classes of hosts need to be considered in epidemiological models—infected and uninfected hosts. This simplification is very helpful for understanding the epidemiology.
Polymorphic allozyme markers are available for nearly all Daphnia species, allowing one to identify multi-locus genotypes very cheaply and quickly (Hebert and Beaton 1993). With an established routine, it is possible to type more than 1000 individuals on a working day, enabling one to follow clone frequencies in replicated populations and monitor microevolutionary changes.
Because experimental studies must compare the treatment group with the control group, the control group must be handled the same way as the treatment group in every respect, except for the experimental factor (parasites, for most purposes in the context of this book).
Here I want to mention a few points about the use and preparation of placebos. A placebo refers to a control treatment that resembles the other treatment in all aspects except the one that is being tested. There are two crucial points to consider when using placebos.
First, the placebo has to resemble the treatment in all factors except the actual treatment factor. Thus, when preparing a spore suspension with macerated tissue from infected Daphnia, the placebo must be prepared with tissue from uninfected Daphnia. Using water as a placebo is not enough, because it differs in more than the absence of transmission stages from the treatment suspension. It may also lack nutritional material for the Daphnia that the treatment suspension may contain.
Second, the placebo may have some effect on the controls. This effect, often called a placebo effect, describes a difference between the placebo treatment and a totally untreated control. If the effect of the actual treatment and the effect of the placebo treatment do not influence each other, this is not a problem, but if the effect of the placebo interacts with the effect of the actual treatment, the results may be difficult to interpret. For example, suppose you test for the immune response of a host after it is exposed to parasitespores. If both the placebo and spore suspensions contain compounds that influence the immune response of the host (e.g., certain bacteria), one obtains estimates of host response, which have to be seen within the light of this suspension. A water control may not have the same effect. The response to the exposure to spores may have been different if the spores had been in a water suspension without any other compounds. I recommend, therefore, using two controls in individual-level experiments: a placebo control and a control without anything. You may not be able to avoid a placebo effect, but it is important to know about it.
Controls have more functions than just being the sample against which the treatment is tested. When testing for the effect of certain treatments on a parasite's performance, infected hosts should be kept under different treatment conditions (e.g., parasite growth under different environmental conditions; transmission rates under different densities). Because all treatment groups are infected, an uninfected control does not seem necessary. There are reasons why uninfected controls (actually placebo-exposed controls) should be included. First, the uninfected controls allow you to verify that all material was uninfected before the start of the experiment. Second, some experiments fail for unknown reasons, e.g., there may be high unexplained mortality. The controls allow you to judge whether the parasites played a role in these results.
In certain experiments, it is not clear whether the treatment applied will show any effect. A negative result is difficult to present in a convincing way, because the nonsignificance of the treatments may have been caused by other reasons than the absence of an effect—the absence of evidence is not evidence for absence. For example, statistical noise may disguise a treatment effect in a poorly executed experiment. To ascertain the quality of the experiment, I recommend using an additional factor that is known to produce a visible effect, even if this effect is not the focus of your research question. For example, one may use two food levels, along with the other treatment. Then if a food effect is apparent, you may convince the observer that other treatment effects could also be found, provided they are there. If you fail to find a food effect, your experiment may have been poorly performed.
A number of Daphnia parasites can easily be bred under laboratory conditions and are therefore suitable for experimental work. These experiments can be conducted in courses on the evolution and ecology of host–parasite interactions but also for research purposes. What follows are some suggestions for simple experiments that will work even if one has little experience with Daphnia parasites.
The transmission stages of horizontally transmitted parasites may be administered to the host in different concentrations. Typically, higher doses are more likely to produce infections (Ebert et al. 2000b; Regoes et al. 2003). To quantify the infection success of parasite isolates, a standardized measure is used: the ID50 (or infective dose 50%), which is the dose at which 50% of the exposed hosts become infected. The ID50 may vary strongly among parasite isolates and host clones (Ebert 1998b). It is usually estimated with a statistical procedure based on infection data (binary data) in response to several different dose treatments.
The success of a parasite also depends on its within-host growth, which in turn depends on within-host competition. The more transmission stages that enter a host, the more competition will occur, thus lowering the success of each individual parasite. In extreme cases (very strong competition resulting from very high doses), the parasite may completely fail to conclude its development (Ebert et al. 2000b). High doses of the parasite may also harm the host more strongly. It has been observed that, with increasing spore dose, host mortality and morbidity increases (Ebert et al. 2000b).
Dose experiments can easily be done with every parasite that is transmitted from a dead host. Transmission stages are collected from dead hosts, and suspensions are produced with different concentrations of spores. Spore concentrations may be varied over several orders of magnitude to observe clear-cut effects. Parasites that are transmitted from living hosts may also be used in dose experiments. For these experiments, one exposes the recipient host to different-sized groups of infected hosts (Ebert 1995). I suggest using at least 10 replicates per dose level to facilitate the statistical analysis.
It is often a challenge to determine the mode of transmission for unknown parasite species. For a course in ecological and evolutionary parasitology, it can be a rewarding exercise to run a series of experiments with selected parasites to determine their mode of transmission. The experiments to test for mode of transmission can be extrapolated from the chapter on transmission. Keep in mind that some parasites can transmit using more than one mode.
Conventionally, one thinks of a parasite as detrimental to the host. However, because it is often difficult to test for the effect of the parasite on its host, there is some belief that many parasites are not harmful. Here I suggest testing for the effect of parasites on the survival and fecundity of individual females. For the simplest type of experiment, a split brood design is useful. In this design, females must be kept under very good conditions so that they produce large clutches of offspring. Females around 15-25 days old produce the largest clutches. Shortly after their release from the brood chamber, offspring should be isolated in individual jars; half of them should be exposed to the parasite, the other half to a placebo. Animals need to be fed daily, and medium must be changed every 3-4 days. The individuals of both treatment groups should be checked daily for survival and offspring production. Detailed descriptions of similar experiments have been published (Ebert 1995; Ebert and Mangin 1997; Bittner et al. 2002).
An alternative to experiments on individuals is to investigate the effect of parasites on their host populations. Such experiments allow the investigator to ask questions that cannot be answered on the individual level, such as: Do infected populations have lower population densities than parasite-free populations? Can parasites drive their host populations to extinction? Do infected populations have more pronounced population size fluctuations? Do hosts/parasites show an evolutionary response to their antagonist?
The beauty of experiments on the population level is that the results relate more closely to the processes in natural systems because they include interactions that arise from the fluctuating numbers of community members, e.g., effects of density-dependent population growth, density dependence of transmission processes, and effects of genetic and demographic (age and size) population structure.
Daphnia and its microparasites compose one of the few systems where both host and parasites have generation times short enough to allow experimental ecological and evolutionary studies to be carried out in real time. The wide range of parasites available allows for the testing and comparison of epidemiological, evolutionary, and genetic models of infectious diseases.
The basic outline of such experiments is rather simple. One can start populations from stock cultures with a mixed age and size distribution. Populations should be large enough to minimize random effects, such as genetic drift, chance extinctions, or large unexplained variation among replicates. Once treatment groups are formed and treatments applied, populations can be followed in regular intervals over long periods of time. It is important to think ahead about the way in which the populations are sampled, because sampling itself may introduce some effect.
The use of experimental epidemiology and evolution as research tools is still not very widespread. Here I introduce a few studies that use these methods in the hope of stimulating more experimental approaches of this type.
Outbreaks of epidemics in vertebrate populations have often been linked to host stress. No similar predictions have been made about the response of invertebrates to stressful conditions. A population-level experiment was designed to test for the effect of food stress on the epidemiology of the gut parasite Glugoides intestinalis (Pulkkinen and Ebert 2004). Infected and uninfected D. magna populations, which had been kept for many generations under a constant high food supply, were exposed to a severe reduction in the amount of available food. Infected and uninfected control populations continued to receive the full amount of food. Changes in parasite and host population size as well as host body length were recorded to determine how the food shortage influenced host and parasite population dynamics. In both infected and uninfected populations, food shortage led to an approximately equal reduction in host density and changes in host body length distribution. Large hosts suffered from higher mortality than smaller hosts, which significantly reduced the mean body length in the starved populations. Because this change was stronger in the infected populations and because large hosts usually carry the most parasites, this change led to a reduction of average parasite spore load and prevalence in the starved populations. These results indicate that food stress for hosts impairs parasite spread in this system and that host mortality can be an important factor in regulating parasite abundance at the population level.
Parasites may influence the competitive ability of their host. This effect can be pronounced when parasites show some degree of specific virulence for otherwise superior competitors. A simple experiment to investigate this effect is to set up populations with two Daphnia species and follow their populations in the presence and absence of a parasite. Bittner (2001) conducted such an experiment with competition between D. galeata and D. hyalina and the parasite Caullerya mesnili. In the presence of C. mesnili, D. hyalina was the superior competitor, whereas it was inferior in the absence of C. mesnili.
Because parasites may alter the competitive ability of certain clones, similar experiments may be done with competition among clones of one Daphnia species. Capaul and Ebert (2003) allowed 21 clones of D. magna to compete in the presence or absence of different parasite species in 10-liter aquaria with a population size of about 1000 animals. The outcome of clonal competition was not only very rapid (strong changes were evident after only 2 months) but also differed among all treatments. A similar design was chosen by Haag (2004), who allowed clones of D. magna to compete under outdoor conditions in mesocosms (rain tons). He also found strong changes in clonal composition that were dependent on time and parasite treatment.
Population-level experiments may also be used to study the evolution of parasites. To test for the effect of host demography on the evolution of parasite virulence, a laboratory experiment was set up in which parasites were allowed to evolve. If the life expectancy of a parasite is short, it is expected to evolve at a higher rate of host exploitation and, therefore, higher virulence, because its penalty for killing the host is minimized. This hypothesis was tested by keeping the horizontally transmitted microsporidian parasite G. intestinalis in monoclonal cultures of D. magna under conditions of high and low host background mortality (Ebert and Mangin 1997; Ebert 1998b). High host mortality and, thus, parasite mortality, was achieved by replacing 70-80% of all hosts in a culture with uninfected hosts from stock cultures every week (replacement lines). In the low mortality treatment, no replacement took place. Contrary to expectations, parasites from the replacement lines evolved a lower within-host growth rate and virulence than parasites from the nonreplacement lines. Across lines, a strong positive correlation between within-host growth rate and virulence was found. The unexpected result was explained by the more severe within-cell competition in the nonreplacement lines, which may have led to selection for accelerated within-host growth. These results point out that single-factor explanations for the evolution of virulence can lead to wrong predictions and that multiple infections are an important factor in virulence evolution.
The Daphnia–parasite system has proved to be a wonderful system for experimental and observational studies, both on the individual and the population levels. In my 15 years of research with this system, I found only two aspects of this system to be lacking, which would make it even more powerful: