Short-term rates of parasite evolution predict the evolution of host diversity


Angus Buckling, Department of Zoology, University of Oxford, South Parks Road, Oxford OX1 3PS. Tel.: +44 (01865) 281305; e-mail:


Coevolution with parasites has been implicated as an important factor driving the evolution of host diversity. Studies to date have focussed on gross effects of parasites: how host diversity differs in the presence vs. absence of parasites. But parasite-imposed selection is likely to show rapid variation through time. It is unclear whether short-term fluctuations in the strength of parasite-imposed selection tend to affect host diversity, because increases in host diversity are likely to be constrained by both the supply of genetic variation and ecological processes. We followed replicate populations of coevolving, initially isogenic, bacteria and phages through time, measuring host diversity (with respect to bacterial colony morphologies), host density and rates of parasite evolution. Both host density and time-lagged rates of parasite evolution were good independent predictors of the magnitude of bacterial within- and between-population diversities. Rapid parasite evolution and low host density decreased host within-population diversity, but increased between-population diversity. This study demonstrates that short-term changes in the rate of parasite evolution can predictably drive patterns of host diversity.


Explaining patterns of genetic and ecological diversities remains one of the biggest challenges in evolutionary ecology. Both theory and experiments suggest parasites are likely to play an important role in the evolution of diversity of their host species, potentially causing both increases and decreases in levels of diversity (Clarke, 1979; Holt & Lawton, 1994; Abrams, 2000; Bohannan & Lenski, 2000; Doebeli & Dieckmann, 2000; Schluter, 2000; Abrams, 2002; Buckling & Rainey, 2002a, b; Chase et al., 2002; Brockhurst et al., 2004; Morgan & Buckling, 2004; Vamosi, 2005). Empirical studies to date have been limited to gross effects of parasites on host diversity, addressing the question of whether evolution of host diversity differs when parasites are present or absent (Bohannan & Lenski, 2000; Buckling & Rainey, 2002a, b; Brockhurst et al., 2004; Morgan & Buckling, 2004; Vamosi, 2005). However, interactions between hosts and parasites are likely to result in host–parasite coevolution (Thompson, 1994), the reciprocal evolution of host defence and parasite counter-defense, which will inevitably result in temporal variation in the strength of parasite-imposed selection.

It is currently unclear if rapid temporal variation (tens rather than hundreds or thousands of generations) in the strength of parasite-imposed selection is important in shaping host diversity (Thompson, 1998). In particular, the ability of host populations to increase in diversity may be limited by the supply of appropriate genetic variation (genetic constraints) and the potential for new variants to successfully invade (ecological constraints, such as Allee effects; Allee 1931). As such, host population diversity may be relatively robust with respect to short-term changes in the strength of parasite-imposed selection. The aim of this study was to determine if temporal variation in parasite-imposed selection, or more specifically the rate of evolution of parasite infectivity, is a good predictor of bacterial diversity in experimentally coevolving populations of bacteria and parasitic bacteriophages. More generally, it is important to understand whether rapid evolutionary change is an important driver of biological diversity over ecological timescales (Thompson, 1998).

Both theory and experiments suggest that parasites can drive both within- and between-population diversities (Buckling & Rainey, 2002a, b). Within-population diversity is likely to be affected via three main processes. Parasites will reduce host diversity if there is convergent (directional) host evolution to resist parasites, but may also promote diversification via competition for enemy-free space, if there exist alternative resistance strategies or trade-offs between competitive ability and resistance to parasites (Holt & Lawton, 1994; Abrams, 2000; Chase et al., 2002; Vamosi, 2005). Parasites can also reduce host diversity via host density reductions, thus decreasing diversifying selection driven by competition for resources (Buckling & Rainey, 2002a, b). Note that this latter effect of parasites on host diversity does not result from direct parasite-imposed selection. Parasites may increase between-population diversity because coevolution is often a stochastic process, and hence will differ between populations (Thompson, 1994; Gandon et al., 1996; Buckling & Rainey, 2002a, b; Morgan et al., 2005). Furthermore, density reductions and genetic bottlenecks caused by parasites may increase the effects of genetic drift, again increasing stochastic differences between populations. In coevolving systems, all these effects on diversity are likely to be most pronounced when parasite-imposed selection on the host (the rate of parasite evolution) is the greatest.

Experimental studies of antagonistic coevolution are made possible by the use of laboratory populations of bacteria and their parasitic viruses (Chao et al., 1977; Levin et al., 1977; Lenski, 1984). We employ the plant-colonizing bacterium Pseudomonas fluorescens SBW25 (Rainey & Bailey, 1996) and a naturally associated viral parasite (phage SBW25φ2) (Buckling & Rainey, 2002a, b). Phages bind to sensitive bacterial cells, inject in their genetic material and replicate using the bacterial cellular machinery. Release of phage particles requires bacterial cell lysis, hence the interaction between bacteria and virulent phage is entirely antagonistic (Lenski, 1984). Bacteria rapidly evolve resistance to the phage, phages subsequently evolve to infect these resistant cells and bacteria evolve resistance again. Unlike other bacteria-phage systems where phage are unable to overcome these ‘second-order’ resistant bacteria (Chao et al., 1977; Levin et al., 1977; Lenski, 1984; Bohannan & Lenski, 2000), up to 10 cycles of bacterial resistance and phage infectivity evolution have been observed during the course of a 400 bacterial generation study (Buckling & Rainey, 2002a, b). Antagonistic coevolution is largely driven by directional selection, with hosts becoming resistant to a wider range of parasite genotypes and parasites infective to a wider range of host genotypes (Buckling & Rainey, 2002a, b).

This system is very useful for studying the impact of coevolution on the evolution and maintenance of host diversity. When propagated in spatially heterogeneous environments (a static glass microcosm containing nutrient-rich medium), P. fluorescens populations rapidly diversify, generating numerous niche specialist genotypes that are readily distinguished by their (heritable) colony morphologies on agar plates (Rainey & Travisano, 1998). Consistent with ecological and population genetics theory (Clarke, 1979), the diversity of these types is maintained by negative frequency-dependent selection. Previous short-term experiments under these experimental conditions demonstrate that phages can both reduce or increase P. fluorescens within-population diversity, depending on abiotic conditions, largely because of phage-imposed density reductions and trade-offs between bacterial resistance and growth rates (Buckling & Rainey, 2002a, b; Brockhurst et al., 2004; Morgan & Buckling, 2004). Between-population diversity is typically increased, regardless of environmental conditions, and this appears to be due simply to different genotypes evolving resistance at different rates in different populations (Buckling & Rainey, 2002a, b; Brockhurst et al., 2004).

Here, we carry out longer term studies of coevolution to determine how well temporal patterns of host diversity can be predicted by the rate of phage infectivity evolution. The rate of phage infectivity can result in changes in diversity as a result of both direct parasite-imposed selection (selection for host resistance) and indirect effects of parasites (changes in host density). Both variables were found to be important predictors of temporal variation in within- and between-population diversities, demonstrating that rapid evolutionary change in parasite populations can play a crucial role in shaping the diversity of their host populations over short-term (ecological) time scales.

Materials and methods

Culturing techniques

Populations were propagated in static 25-mL glass universal bottles (microcosms) containing 6 mL of King's Medium B (KB) in a static 28 °C incubator. Twenty-four replicate microcosms were inoculated with 108 cells of P. fluorescens isolate SBW25 (grown for 24 h at 28 °C in an orbital shaker at 0.45 g), and to 12 of these were added 105 clonal particles of a naturally associated DNA phage, SBW25φ2. Sixty microlitres of each culture was transferred to fresh medium every 2 days, for 30 transfers (approximately 220 bacterial generations). Note that the ancestral bacteria are sensitive to the phage. Cultures were frozen at −86 °C in 20% glycerol at every second transfer. Bacterial densities were estimated by plating diluted cultures onto standard KB agar, and counting the number of colony-forming units (CFUs) after 48 h of incubation at 28 °C. Phage densities were determined from the number of plaque-forming units produced after 24 h growth at 28 °C on soft agar containing exponentially growing ancestral bacteria; phage populations were isolated from the cultures by centrifuging cultures with 10% chloroform, which lysed and pelleted bacterial debris (Buckling & Rainey, 2002a, b).

Measurement of diversity

Diversity was measured by determining the morphologies of 100 random colonies on KB agar plates (Rainey & Travisano, 1998). Within-population diversity was calculated as the complement of Simpson's index of concentration (1 − λ) (Simpson, 1949), with a correction for any variation in the number of colonies measured (Lande, 1996):


where pi is the proportion of the ith morph and N is the total number of colonies sampled. This measure is the probability that two randomly selected morphotypes are different. Between-population diversity was calculated as a variance (d2) (Lande, 1996): the probability that two types, randomly chosen from any population, are different:


where inline image is the mean proportion of the ith morph across all j populations.

Measuring bacterial resistance and phage infectivity

Coevolutionary arms races are characterized by the continual evolution of host defence and parasite counter-defence. For simplicity, we consider only the binary traits of bacterial resistance/sensitivity and phage infectivity/noninfectivity. The resistance of a particular bacterial population to a particular phage population (or the infectivity of a phage population to a bacterial population) was determined by streaking 20 independent bacterial colonies across a perpendicular line of phage that had previously been streaked on a KB agar plate. A colony was defined as resistant if there was no inhibition of growth, otherwise it was defined as sensitive. Ancestral SBW25 was used on every plate as a control (Buckling & Rainey, 2002a, b).

Measurement of phage evolution

Selection on phage infectivity shows a time lag, such that contemporary phage should be better able, than phage from the immediate past, to infect bacteria from the immediate past. Put another way, contemporary bacteria were expected to be more resistant to contemporary phage than bacteria from the immediate past. To measure this, we determined, at every second transfer (approximately 15 bacterial generations), the resistance of bacterial populations from two transfers in the past to their contemporary phage and to the current phage population (Buckling & Rainey, 2002a, b). We define the rate of phage evolution at time point t as the change in phage infectivity from time point t − 2 to time point t, to bacteria from time point t − 2. We simultaneously measured bacterial resistance to the ancestral phage, and the ability of phage populations to infect the ancestral bacteria.

Statistical analyses

We analysed how the rate of parasite evolution and bacterial density affected bacterial diversity through time as linear (mixed) models in R, with significance of terms determined by removal of terms from the model. We considered a range of time lags on these two covariates: the rate of infectivity evolution at four transfers in the past, and contemporary density, were the best predictors of contemporary diversity and diversification.

We first considered how these covariates affected bacterial diversity through time. As between-population diversity is by definition a measure across all 12 populations, we also calculated the mean values across all 12 populations for within-population diversity (mean of 1 − λ) and both covariates. To de-trend changes in bacterial diversity through time, we subtracted mean diversity in the presence of phage (12 replicate populations) from mean diversity in the absence of phage (12 control populations) at each time point. We then used multiple regression to test the effect of the rate of phage evolution of infectivity, and current bacterial density, on detrended, between-population diversity. For comparison, we calculated the average within-population diversity (mean of 1 − λ), detrended it by subtracting mean diversity in the absence of phage at each time point, then regressed this response variable against mean rates of phage evolution and mean bacterial densities.

We next addressed how rates of parasite evolution and host density affected changes in diversity through time. Analyses using means of all replicates, although essential for between-population diversity, lost a considerable amount of within-population information. We therefore carried out an analysis of within-population rates of diversification, using REML (in R) to account for the longitudinal nature of the data: each coevolving pair of populations was measured repeatedly through time. This analysis also allowed us to consider the autocorrelated influence of recent bacterial diversity on rates of diversification. We used the change in bacterial diversity [(1−λ)t−(1−λ)t−1] as our response variable and multiply regressed it against bacterial diversity at the previous transfer, rate of evolution of phage infectivity over the previous two transfers, and current bacterial density. Replicate population was included as a random effect. All statistical models were checked for homoscedasticity and normality of residuals.

Results and discussion

We followed the rate of parasite evolution, bacterial density and bacterial diversity in coevolving populations of bacteria and phages through time. Coevolution is apparent in this study from the increase in infectivity (on contemporary bacteria) between contemporary and future phage populations, and the subsequent increase in resistance (to contemporary phage populations) from past to contemporary bacterial populations (Fig. 1a) (Buckling & Rainey, 2002a, b). Consistent with previous short-term work, we found that populations coevolving with phages showed decreased density (Fig. 1b; mean through time: t22 = 14.95, P < 0.001), decreased within-population diversity (Fig. 1c; mean through time: t22 = 11.07, P < 0.001) and increased between-population diversity (Fig. 1d; diversity higher in the presence of phage for 15 of 18 time points; Sign test: P < 0.01), relative to bacterial populations evolving in isolation.

Figure 1.

 Dynamics of coevolution, density and diversity. The mean (−1 SEM) proportion of bacteria resistant to ancestral phage (open squares), contemporary phage (closed circles) and phage from two transfers in future (open circles). The vertical arrow indicates the increase in phage infectivity over two transfers (contemporary to future phages) and the diagonal arrow indicates the subsequent increase in bacterial resistance (past to contemporary) (a). Mean (±SEM) densities of bacteria evolving in the absence of phages (closed circle) and in the presence of phage (open circle) and density of phages (closed triangles) (b). Mean (±SEM) within-population morphological diversity of bacteria evolving in the absence (closed circles) and presence of phages (open circles) (c). Between-population morphological diversity of bacteria evolving in the absence (closed circles) and presence of phages (open circles) (d).

We wanted to determine if the rate of phage evolution (direct parasite-imposed selection) and bacterial density (indirect parasite-imposed selection) are important predictors of bacterial diversity. We determined the effect of these two covariates, with various time lags, using different models for both between-population and average within-population diversities. In the case of between-population diversity, both recent phage evolution (four transfers in the past) and contemporary bacterial density were significant predictors of bacterial diversity (Fig. 2a; rate of phage evolution, F1,6 = 13.682, P = 0.010; bacterial density, F1,6 = 9.964, P = 0.020; between-population diversity increased with increasing mean bacterial density and with increasing rates of phage evolution). In the case of mean within-population diversity, bacterial diversity was not influenced by contemporary bacterial density (Fig. 2b; F1,6 = 1.981, P = 0.209) but decreased with increasing rate of phage evolution (F1,7 = 118.210, P < 0.001). These data strongly suggest that the rate of phage evolution is a crucial driver of the evolution of host diversity but that the direction of its effect differs when considering within- or between-population diversity.

Figure 2.

 The difference in diversity between bacteria evolving in the presence and absence of phage (presence − absence) as a function of the rate of parasite evolution through time. Mean difference in between-population diversity (a); difference in within-population diversity (b). Fitted lines represent regression models; in (a) the line is fitted using the mean value of bacterial density.

Analysis of the non-averaged data suggested that within-population diversity tended to decrease between transfers at high rates of phage evolution and increase at low rates (F1,69=21.765, P < 0.001; Fig 3a). Conversely, bacterial diversity tended to increase between transfers at high density and decrease at low density (F1,69=5.525, P=0.022; Fig. 3b). There was evidence of a negative feedback between recent diversity and the rate of diversification: change in diversity decreased with increasing diversity (F1,69 = 75.545, P < 0.001; Fig. 3c). These data demonstrate that the rate of phage evolution determines the magnitude and direction of short-term changes in within-population diversity through time.

Figure 3.

 The influence of rate of phage evolution, bacterial density and recent bacterial diversity on change in bacterial diversity through time. Scatter plots show all data irrespective of replicate population, but fitted lines show modelled relationships for an average population with mean values for all covariates except that displayed on the x-axis. Change in diversity (Simpson's index at time t minus Simpson's index at time t − 1) plotted against (a) rate of phage evolution (from t − 2 to t); (b) bacterial density at time t; (c) bacterial diversity at time t − 1.

An important implication of this work is that effects of parasite evolution are transient: bacterial populations rapidly returned to phage-free levels of diversity when little coevolution occurred. This appears inconsistent with previous experimental evolution studies of P. fluorescens in the absence of phage (Buckling et al., 2003; Brockhurst et al., 2007). In the present study, phages tend to reduce within-population diversity, but this does not appear to greatly affect the ability of the population to subsequently increase in diversity. However, in other phage-free experiments, experimental manipulation of within-population diversity constrained the subsequent evolution of diversity. In one study (Buckling et al., 2003), the continual propagation of single bacterial clones specializing in the broth phase resulted in large fitness improvements in these clones, but a greatly reduced ability to diversify. This was presumably because novel genotypes that were able to occupy alternative niches were still at a fitness disadvantage relative to highly specialized genotype from which they evolved. In the second study (Brockhurst et al., 2007), the total diversity that evolved in microcosms was found to be reduced in the presence vs. the absence of single resident genotypes that had been previously been established at high frequencies. The reduction in total diversity was brought about by the resident genotypes preventing the invasion of multiple genotypes that would otherwise have coexisted in the resident's niche. The apparent inconsistency between these studies and the present study is probably because phage-imposed bottlenecks are less extreme than the experimental bottlenecks; and that the continually changing selection pressures resulting from coevolution with phage may indirectly select for elevated bacterial mutation rates (in the same way that coevolution may favour sexual reproduction; West et al., 1999), reducing possible constraints on diversification.

These data provide some important insights into the mechanisms responsible for phage-mediated changes in host diversity, most notably that both the rate of evolution of parasite infectivity and parasite-imposed reductions in bacterial density are important predictors. With respect to within-population diversity, phage-mediated reductions in density presumably reduce resource competition-mediated diversifying selection to occupy novel niches (Buckling & Rainey, 2002a, b). Increasing rates of phage evolution will impose increasingly strong directional selection for phage resistance, purging genetic diversity. Our previous short-term manipulative studies did not detect the importance of selection for host resistance, and only identified parasite-imposed reductions in host density as an important driver of within-population diversity (Buckling & Rainey, 2002a, b). In the case of between-population selection, selection for phage resistance is likely to increase diversity because different genotypes evolve resistance at different rates, and hence dominate, in different populations. This divergence will increase with increasing strength of selection for resistance. Diversity is further increased by reduced density, presumably because evolutionary stochasticity is further increased with decreasing population sizes. Our previous short-term studies were only powerful enough to identify parasite infectivity, but not density, as an important driver of between-population diversity (Buckling & Rainey, 2002a, b). We believe the conclusions of the current work are more robust because our previous study considered diversity at a single time point at an early stage of coevolution, and involved an artificial manipulation of density in the absence of phages.

In summary, we have shown that temporal patterns of host diversity are predictably driven by rates of parasite infectivity evolution: high rates of evolution results in time-lagged reductions in within-population diversity and increases in between-population diversity. These effects are further increased by parasite-mediated reductions in host density. Crucially, these results demonstrate that rapid parasite evolution can predictably drive changes in host diversity over ecological timescales. This work complements previous studies demonstrating that rapid evolution can drive the population dynamics of interacting species (Bohannan & Lenski, 2000; Yoshida et al., 2003).


We thank the Leverhulme Trust, the Royal Society and the European Social Fund for funding. Suggestions from Mike Ritchie, Mike Brockhurst, Nick Colegrave and two anonymous referees improved the manuscript.