The ecological forces shaping adaptive radiations are of great interest to evolutionary ecologists. Here, we experimentally test the hypothesis that the diversification of a lineage should be limited in the presence of competition from another taxon. We do this by studying a model microbial adaptive radiation (the generation of phenotypic diversity in asexual lineages of the bacterium Pseudomonas fluorescens) in the presence or absence of a competitor (Pseudomonas putida). In a spatially heterogeneous environment, the competitor P. putida reduced P. fluorescens population size only slightly and had no effect on diversification. In a spatially homogeneous environment, the competitor reduced P. fuoresecens population size to a much greater extent. Again the final extent of diversification in P. fluorescens was not affected by the competitor, but early diversification was accelerated. In this environment, P. putida suppressed the growth of a common variant of P. fluorescens and directly or indirectly facilitated the growth of a rare morph. Our results suggest that competition experienced by diversifying lineages may have complex effects on adaptive radiations not fully captured by current theory.
Resource competition is perhaps the chief process determining the form of adaptive radiations. It has generally been thought that competition within a radiating lineage should promote diversification, whereas competition from other taxa should inhibit it by reducing the number and width of potential niches (Mayr 1942; Lack 1947; Simpson 1953; Schluter 2000). Empirical support for the hypothesis that competitors impede adaptive radiation has mainly come from comparative studies of morphological divergence and speciation rates of sister taxa on islands (where unrelated competitors are absent) versus continents (where unrelated competitors are present; Schluter 1988; DeSalle 1995). However, the assumption that competing taxa limit the number of niches available for diversifying lineages may not be met. Competitors can create novel niches by increasing the complexity of abiotic and biotic environments, for example, by modifying the physical structure of ecosystems (Jones et al. 1997; Palmer and Maurer 1997; Vellend and Geber 2005), and thus the presence of competing taxa may even promote diversification (Emerson and Kolm 2005). The latter “diversity drives diversification” hypothesis has been supported by a positive correlation between species diversity and the rate of endemic diversification in certain plant and arthropod groups across archipelagos (Emerson and Kolm 2005). By their nature, adaptive radiations are difficult to study experimentally and hence most studies have so far employed comparative approaches (but see Brockhurst et al. 2007; Fukami et al. 2007).
Here, we test experimentally whether a potential interspecific competitor affects diversification in a model microbial system. In a spatially heterogeneous environment (static tubes containing growth media), initially isogenic populations of the bacterium Pseudomonas fluorescens rapidly diversify into numerous phenotypes that can be distinguished by heritable colony morphology as well as by the ecological spatial niches they occupy. The phenotypes fall into three categories: “smooth morphs” (SM) that resemble the ancestral type and inhabit the liquid phase, “wrinkly spreaders” (WS) that form a biofilm at the air-broth interface, and “fuzzy spreaders” (FS) that colonize the bottom of tubes (Rainey and Travisano 1998; Buckling et al. 2000; Kassen et al. 2000). Each category consists of several phenotypes that can be distinguished on the basis of heritable difference in colony morphology, for example, small-WS, large-WS, and wheel-like WS types within the WS category. Phenotypic diversity in this system is maintained through negative frequency-dependent selection (Rainey and Travisano 1998; MacLean et al. 2005) as well as stochastic persistence of morphs with near-equal fitness (Fukami et al. 2007; Zhang et al. 2009). Populations in a spatially homogeneous environment (continuously shaken microcosms) are dominated by SM phenotypes (Rainey and Travisano 1998; Buckling et al. 2000; Kassen et al. 2000). It has been shown that in this system the initial presence of niche specialists (such as the biofilm-forming WS types) can suppress the diversification of the ancestral type (Brockhurst et al. 2007; Fukami et al. 2007), an example of niche occupation limiting the extent of adaptive diversification. Here, we address whether the initial presence of a congeneric competitor, the bacterium Pseudomonas putida, limits the diversification of P. fluorescens.
Materials and Methods
Pseudomonas putida was used as the competitor species in this study. This species is likely to cooccur in the same microhabitats as P. fluorescens as both species inhabit similar soil environments, are often associated with the same species of plants, and grow optimally at 25–30°C (Ramos 2004). We grew P. fluorescens SBW25EeZY6KX (Bailey et al. 1995) and P. putida UWC1::GFP (van Overbeek et al. 2002) in monocultures and mixtures, in both unshaken (spatially heterogeneous) and shaken (spatially homogeneous) microcosms. Six replicate microcosms of each species combination were destructively sampled every two days until day 14 when diversification would have run its course (Rainey and Travisano 1998). Therefore, we established 252 microcosms (three species compositions × two environments × seven points of sampling time × six replicates).
Bacteria were grown in 30 mL universal tubes with loose lids, each containing 6 mL of King's Medium B (KB). Each microcosm was initialized with 107 cells per species. The unshaken microcosms were cultured at 28°C in a static incubator, and the shaken microcosms were grown in a shaking incubator (150 rpm, shaking radius 25 mm) at 28°C. Bacterial densities were measured by plating diluted cultures onto KB agar plates containing X-gal and counting the number of colony forming units (CFUs) after 72 h of incubation at 28°C. The colonies of P. fluorescens SBW25EeZY6KX (with a lacZY insert) have a blue color and are easily distinguished from those of P. putida UWC1::GFP. During the experiment, we observed no diversification in P. putida in terms of colony morphology or niche occupancy; in particular, P. putida never formed a biofilm at the air-broth interface.
We measured the total number of morphologically distinct phenotypes (phenotypic richness) in P. fluorescens based on a sample of approximately 100 randomly chosen colonies (Buckling et al. 2000; Kassen et al. 2000). Diversity was expressed as the complement of Simpson's index (Simpson 1949), where pi is the frequency of the ith phenotype. The data were analyzed using analyses of variance (ANOVA) with day and the presence or absence of the competitor as categorical explanatory variables. Log transformations were applied to bacterial density and phenotypic richness data, and arcsine transformation to proportional data.
In the spatially heterogeneous microcosms, P. fluorescence population size declined over the 14-day course of the experiment with the presence of the competitor P. putida leading to an on average 13% reduction in P. fluorescens population size (Fig. 1A; competitor, F1, 70= 5.93, P= 0.017; day, F6, 70= 36.93, P < 0.001; interaction, F6, 70= 2.59, P= 0.025). Over time the phenotypic richness of P. fluorescens increased rapidly toward a plateau and were unaffected by the presence of the competitor species (Fig. 1C; competitor, F1, 70= 0.15, P= 0.696; day, F6, 70= 106.05, P < 0.001; interaction, F6, 70= 1.18, P= 0.329). Phenotypic diversity (1-λ) also increased over time and was greater in the presence of the competitor species in the early (but not the later) part of the experiment (Fig. 1E; competitor, F1, 70= 13.59, P < 0.001; day, F6, 70= 155.63, P < 0.001; interaction, F6, 70= 2.55, P= 0.027). This effect is most likely explained by the presence of P. putida increasing diversity within SM phenotypes early in the experiment (Figs. S1 and S2 show the relative ratios of the three categories of phenotypes, phenotypic richness, and diversity within each category, in the presence and absence of the competitor species).
In the homogeneous microcosms, the presence of the competitor P. putida led to a reduction in the population size of P. fluorescens by 73% on average, and P. fluorescens population size also decreased over time (Fig. 1B; competitor, F1, 70= 255.10, P < 0.001, day, F6, 70= 42.27, P < 0.001, interaction, F6, 70= 29.78, P < 0.001). The phenotypic richness of P. fluorescens increased over time and was higher in the presence of the competitor in the early but not the late parts of the experiment (Fig. 1D; competitor, F1, 70= 0.29, P= 0.594; day, F6, 70= 4.73, P < 0.001; interaction, F6, 70= 2.35, P= 0.039). Phenotypic diversity was higher in the presence of the competitor and did not show a simple trend with time (Fig. 1F; competitor, F1, 70= 53.49, P < 0.001; day, F6, 70= 4.94, P < 0.001; interaction, F6, 70= 5.31, P < 0.001).
In the homogeneous microcosms, P. fluorescens populations contained two SM variants (a common variant with large colonies and a rare variant with small colonies, referred to below as “large-SM” and “small-SM,” respectively) and very rare WS types (<4%). The two SM variants showed different responses to the presence of P. putida. The density of “large-SM” was reduced in the presence of the competitor, particularly in the early stages of the experiment (Fig. 2; competitor, F1, 70= 29.13, P < 0.001; day, F6, 70= 4.79, P < 0.001; interaction, F6, 70= 4.76, P < 0.001), whereas that of “small-SM” increased, particularly at the beginning (Fig. 2; competitor, F1, 70= 5.85, P= 0.018; day, F6, 70= 4.44, P < 0.001; interaction, F6, 70= 3.68, P= 0.003).
In designing the experiments, we assumed P. putida would be an interspecific competitor acting to affect the ecological diversification of P. fluorescens. However, we found that the competitive effect exerted by P. putida was dependent on environmental context: P. fluorescens population size was weakly reduced in heterogeneous microcosms, but there was a significant reduction in homogeneous microcosms (Fig. 1). By contrast, the presence of P. fluorescens strongly reduced the population size of P. putida in both heterogeneous and homogeneous microcosms (Fig. S3). A possible explanation for this is that P. putida is less tolerant of oxygen-poor environments than P. fluorescens (Mahendran et al. 2006). In shaken microcosms, where oxygen availability is high, the two species had reciprocal negative effects on each other. In static microcosms, where oxygen is probably a limiting factor (Kassen et al. 2004), the population size of P. putida was always greatly reduced by the presence of P. fluorescens, whereas the reverse did not occur.
In both heterogeneous and homogeneous environments, we did not detect a significant effect of P. putida on the extent of the adaptive radiation in P. fluorescens. By the end of the experiment (day 14), phenotypic richness in P. fluorescens did not differ between microcosms with and without the competitor in either the heterogeneous (nonshaken) or homogeneous (shaken) environments (Fig. 1), suggesting that the competitor species did not have an impact on the number of niches available to P. fluorescens, regardless of the intensity of competition. This does not support the suggestion that competition from other taxa should impede the diversification of a lineage (Mayr 1942; Lack 1947; Simpson 1953).
The early stage of diversification in the homogeneous microcosms (where two SM variants and a rare WS form evolved from the ancestral SM-like type) occurred faster in the presence of the competitor (Fig. 1) suggesting that competitors can alter the tempo of an adaptive radiation (Emerson and Kolm 2005). However, this is unlikely to be explained by competitors creating novel niches, as the final extent of diversification was not affected by the presence of the competitor. A more parsimonious explanation is that the competitor competed strongly with the ancestral P. fluorescens phenotype and so further increased the fitness of derived phenotypes. We observed different responses of phenotypic variants in the diversifying P. fluorescens lineages to the presence of the competitor. The common and ancestral phenotype in P. fluorescens, “large-SM,” was negatively impacted by the presence of P. putida, but a rare type, “small-SM,” was initially favored (Fig. 2). This may likely be explained by that P. putida indirectly promoted the growth of small-SM by suppressing their common competitor, large-SM (Scheffer and van Nes 2006), although it is also possible that the competitor directly stimulated small-SM growth, for example by secreting substances that can be metabolized by this morph.
Our results suggest that biotic interactions such as competition may influence the patterns of diversification observed in different taxa, including multicellular eukaryotes. For instance, several adaptive radiations exhibit “overshooting” dynamics, where species numbers increase rapidly in the initial stages and then decline slowly (Gillespie 2004; Gavrilets and Vose 2005; McPeek 2008). This has also been observed in the phenotypic diversification of P. fluorescens in static microcosms (Fukami et al. 2007; Meyer et al. 2011). Overshooting is normally explained as the result of the interaction between intrinsic speciation and extinction rates (Gavrilets and Vose 2005; McPeek 2008) without the involvement of biotic interactions. Our findings suggest that the role of competitors in early diversification deserves greater attention.
In summary, our work with a laboratory microbial model of adaptive radiation suggests that competition from taxa outside a radiating lineage may not limit diversification and may even accelerate diversification if the competitor only affects the most abundant variants and prevents them from becoming dominant. The generation of microbial diversity in asexual lineages of bacteria is clearly not exactly analogous to evolutionary divergence in multicellular organisms occurring on geological time scales. Nevertheless, such model systems have provided many valuable insights into ecological processes that are otherwise impossible to study experimentally (Buckling et al. 2009). Our results suggest that competition can affect the course of adaptive radiation, although not in the simple ways suggested by current theory.
Associate Editor: S. Nuismer
This work was funded by Natural Environmental Research Council UK through NERC Centre for Population Biology at Imperial College London, and QGZ was supported by the National Natural Science Foundation of China (31070379, 31030014, and 31121003).