Experimental multi-species microbial (co)evolution results in local maladaptation

Interspecific coevolutionary interactions can result in rapid biotic adaptation, but most studies have focused only on species pairs. Here, we (co)evolved five microbial species in replicate polycultures and monocultures and quantified local adaptation. Specifically, growth rate assays were used to determine adaptations of each species’ populations to (1) the presence of the other four species in general and (2) sympatric vs allopatric communities. We found no evidence for general biotic adaptation: ancestral, polyculture- and monoculture-evolved populations did not have significantly different growth rates when measured within communities. However, 4/5 species’ growth rates were significantly lower within the community they evolved in. This “local maladaptation” suggests that species evolved increased competitive interactions to sympatric species’ populations. This increased competition did not affect community stability or productivity. Our results suggest that (co)evolution within communities can increase competitive interactions that are specific to (co)evolved community members.

The primary aim of this study was to determine whether or not there is a tendency for 99 populations to become locally adapted following evolution in multi-species communities: i.e. if 100 they are better adapted to the community they evolved in compared to novel communities. To this 101 end, we experimentally (co)evolved soil bacteria species, propagated as replicate polycultures in 102 nutrient media in which they stably coexist (Padfield et al. bioRxiv). We compared the magnitude 103 of local adaptation to adaptation to biotic versus abiotic conditions per se by also evolving the

128
Where multiple morphotypes had evolved, only the most common morph was isolated.

129
Growth rate assays 130 Using isolated clones, we conducted a series of growth rate assays to assess adaptation ( Figure 1).

131
Each polyculture line was grouped into blocks with one monoculture line of each species (i.e. one    against interacting fixed effects of evolutionary history (ancestor and polyculture-or monoculture-evolved population) and species identity. For the mixed effects model testing biotic adaptation, a 173 random effect of "block" was included to account for which of the assembled communities growth 174 rates had been measured. Local adaptation was tested using a linear mixed effects model in which 175 growth rate was tested against community type (sympatric/allopatric) with species identity as an 176 interacting factor. A random effect of "clone" was included to account for the paired experimental 177 design. 178 We estimated the ability of each species' population invade from rare by conducting one 179 sample t-tests on each focal species within each combination with the null hypothesis being that 180 relative invader fitness is equal to 1. A relative invader fitness greater than 1 would indicate that 181 the invading species has a higher growth rate than the other community members, thus its fitness  All data was analysed in R (v 3.5.1) (Team 2013) and all plots were made using the R 196 package 'ggplot2' (Wickham 2016). Four treatment replicates which became contaminated were 197 removed from analyses (Table S1). Linear mixed effect models were fitted using the R package between species ( 4 2 = 175.9, p < 0.001; Figure 2a).

215
To estimate abiotic adaptation, we measured growth of the same clones as above in 216 monoculture. The growth rate of polyculture-evolved clones did not differ from their respective 217 ancestors, and only two monoculture-evolved populations showed evidence of abiotic adaptation.

218
While there was a significant interaction between species identity and evolutionary history (F8, 31.5 219 = 3.64, p < 0.001), comparisons between ancestral, monoculture-and polyculture-evolve 220 populations showed that this was driven only by populations evolved in monoculture (Figure 2b).

249
Community level properties 250 We have previously shown that all five species can stably coexist (i.e. each can invade from rare)  Figure 4). 257 We also assessed whether evolutionary history impacted community productivity. As local 258 adaptation assays showed species growth rates were greater in allopatric communities relative to 259 sympatric communities, we determined if randomly assembled communities from allopatric 260 polycultures achieved higher total densities than sympatric and monoculture-assembled 261 communities. Overall, there was no significant effect of evolution on productivity between 262 communities composed of monoculture-or polyculture-evolved (sympatric/allopatric) clones 263 (F2,33 = 1.9, p = 0.165; Figure 5a). However, the structure (species proportions) and consequently the relative contribution of each species to productivity varied between communities, as apparent 265 from a significant interaction between species identity and community assembly ( 8 2 = 1186.2, p < 266 0.001; Figure 5b). Significant pairwise comparisons are summarised in Table S5.