DISCUSSION
Migration of individuals from high- to low-quality habitats often help to maintain population sizes in the latter and may have varying effects on local adaptation. The high-quality habitats may also support greater abundances of natural enemies, and the effects of source-sink dynamics on local adaptation in low-quality habitats might fundamentally be altered by coevolving enemies. Our study suggests a negative effect of immigration of coevolving enemies on abiotic local adaptation of victims, particularly when enemies, but not victims themselves, migrate from source to sink habitats.
While bacterial evolution lines in a low temperature habitat showed obvious evolutionary adaptation in terms of population growth performance, bacterial immigration from an optimal temperature habitat reduced such adaptation (B versus B+IB microcosms, Figure 3), consistent with a “swamp by gene flow” view (Bennett & Lenski 2007; Rodríguez-Verdugo et al. 2014; Micheletti & Storfer 2020). It is likely that mutational supply was not, or only weakly, limited in our bacterial populations at 10°C (bottleneck population size > 3 × 107), and immigration had little positive effect on mutation supply, but affected the fixation of beneficial mutations.
The presence of phages limited abiotic adaptation of bacterial populations. It is possible that fixation of phage-resistant mutations may constrain the acquisition of beneficial mutations for abiotic adaptation (Scanlan et al. 2015). Furthermore, the fact that our bacterial growth performance may become even poorer than the ancestral strain (in BP+IP coevolution lines) suggests fitness costs of bacterial resistance evolution, consistent with earlier studies (Buckling & Rainey 2002; Brockhurst et al. 2004; Buckling et al.2006). The bacterial growth performance from BP and BP+IBP coevolution lines was not poorer than the ancestral strain, though no difference in bacterial resistance to phages was found among the three types of coevolution lines. This suggests that bacterial adaptation that compensated for the fitness costs of resistance should have occurred; and such abiotic adaptation was less efficient in the BP+IP microcosms (probably due to the much lower bacterial population sizes and thus lower mutation supply). For the BP+IBP coevolution lines, the bacterial immigrants per se may have promoted bacteria abiotic adaptation, contrary to the effect of bacterial immigrants to the B+IB microcosms.
The negative impact of immigrant phages on bacterial sink adaptation is presumably because the source phages evolved greater infectivity than sink phages and/or achieved higher densities. While the intensity of bacteria-phage coevolution, and hence the extent of phage infectivity evolution, did not differ between our 28 and 10°C microcosms at transfer 20, weak signals of more intense coevolution were detected for the core microcosms at transfer 8 (Supplementary text; Figure S2). This is consistent with previous studies: A previous short term (6-transfer) experiment showed that a modestly low temperature (15°C) could slow bacteria-phage coevolution compared with a higher temperature (25°C) (Zhang & Buckling 2016); and a 10-transfer coevolution study across 3 temperatures (8, 17, 28°C) found more intense coevolution at higher temperatures (Gorter et al. 2016). We suggest that the extremely low bacterial population sizes of BP+IP microcosms during the early stage of evolution experiment result from a “mass effect” of phage immigrations (and higher phage infectivity if coevolution was faster in 28°C microcosms at the early stages), and the low population sizes limited abiotic adaptation. The persistently lower bacterial population sizes of BP+IP microcosms compared with BP microcosms at the late stages of the evolution experiment, however, might be due to the poorer growth performance as a consequence of limited abiotic adaptation, but not greater top-down control effect by phages. The top-down control effect at the end of the evolution experiment could be estimated for BP and BP+IP microcosms by comparing their population sizes (in the presence of phages) and the abiotic adaptation measures (in the absence of phages); and this suggest that phages reduced bacterial population densities by 73% in BP and 62% in BP+IP microcosms (Figure S3).
Migration in coevolving systems often have asymmetric effects on victim and enemy populations (Gandon et al. 1996; Lion et al.2006). For instance, under homogeneous landscapes, phage Φ2 usually benefits from migration more than P. fluorescens on a global scale as phages are more genetically constrained and have a lower evolutionary potential (Brockhurst et al. 2003, 2007; Morganet al. 2005, 2007). Under heterogeneous landscapes, mixed evidence was found for the effect of gene flow on the evolution of phage infectivity (Forde et al. 2004, 2007); and a net beneficial effect of bacterial immigrations on abiotic adaptation was inferred in the BP+IBP treatment here. Therefore, how migration affects the coevolving populations may depend on not only the heterogeneity between patches, but also the migration pattern of victims and enemies (Gandon & Michalakis 2002; Tigano & Friesen 2016; Zhang & Buckling 2016).
In summary, by (co)evolving bacteria and phage populations under various migration regimes across heterogenous environments (Figure 1), we found that abiotic adaptation of bacteria in sink environment was limited by the immigrated phages from the source environment. These results highlight that traditional thinking about source-sink dynamic should be interpreted with caution in antagonistic coevolution systems; immigrant victims and enemies may have opposing effects on the abiotic adaptation of victims in the edge habitats. Future work involving a variety of stress gradients would determine the generality of these findings, and therefore help to understand species’ range expansion in nature with implications for conservation planning (Engelkes et al. 2008; Benning & Moeller 2021).