In a nutshell, a reciprocal transplant experiment reveals local adaptation and fitness trade-offs in response to urban evolution in an acorn-dwelling ant

Abstract

Urban-driven evolution is widely evident, but whether these changes confer fitness benefits and thus represent adaptive urban evolution is less clear. We performed a multiyear field reciprocal transplant experiment of acorn-dwelling ants across urban and rural environments. Fitness responses were consistent with local adaptation: we found a survival advantage of the "home" and "local" treatments compared to "away" and "foreign" treatments. Seasonal bias in survival was consistent with evolutionary patterns of gains and losses in thermal tolerance traits across the urbanization gradient. Rural ants in the urban environment were more vulnerable in the summer, putatively due to low heat tolerance, and urban ants in the rural environment were more vulnerable in winter, putatively due to an evolved loss of cold tolerance. The results for fitness via fecundity were also generally consistent with local adaptation, if somewhat more complex. Urban-origin ants produced more alates in their home versus away environment, and rural-origin ants had a local advantage in the rural environment. Overall, the magnitude of local adaptation was lower for urban ants in the novel urban environment compared with rural ants adapted to the ancestral rural environment, adding further evidence that species might not keep pace with anthropogenic change.

Keywords: Global change; heat island; natural selection; thermal physiology; urbanization.

Figure 1

Fitness trade‐offs for the final four census points (census 8 through 11) (top row), and the monthly temperature profile with all census points (census 1 through 11) indicated (bottom row). Fitness trade‐off plots include estimated mean survival ± 1 SE for each combination of source population and environment from a generalized linear model that accounts for temporal autocorrelation. Note that census 8 y‐axis is scaled from 0.825 to 1, whereas censuses 9–11 are all scaled from 0.3 to 0.825 owing to high overwinter mortality following census 8. On the temperature profile plot, arrows indicate the timing of each census (at the monthly resolution); black arrows correspond with the census points highlighted in the top row. Red triangles indicate a warm‐season census (predominantly summer into fall) and blue asterisks indicate a cool‐season census (predominantly winter into spring), which correspond with the designations used in the analysis of seasonal bias in between‐census survival (Fig. 3). These symbols refer to the period between the previous and focal census point, for example, the red triangle at census 3 refers to a warm period between census 2 and 3, whereas the blue asterisk at census 4 refers to a cool period between census 3 and 4. Temperature indices include average monthly temperature (T _avg), the extreme monthly minimum and maximum temperatures (T _emnt and T _emxt), and the average monthly minimum and maximum temperatures (T _min and T _max). Dates are reported as year‐month.

Figure 2

Survival curves (cumulative) over the entire experiment for each combination of source population and environment. Predicted values from a generalized linear model that accounts for temporal autocorrelation.

Figure 3

Bias in the seasonal timing of colony survival. Ratio ± 1 SE of alternative/own treatment between‐census survival as a function of the summer versus winter season. Results are presented separately for rural‐origin and urban‐origin treatments. The dashed gray line indicates equal survival between away and home treatments; below the line indicates a survival advantage of the home treatment, and above the line indicates a survival advantage of the away treatment.

Figure 4

Estimated number of alates (reproductive ants) ± 1 SE produced per colony over the course of the transplant experiment for each combination of source population and environment. Statistical significance of each pairwise contrast of the two source populations within a single environment and of the two environments for a single source population is indicated with brackets and values (NS = nonsignificant at the P < 0.05 level).