Thermal adaptation in a holobiont accompanied by phenotypic changes in an endosymbiont

Abstract

How and if organisms can adapt to changing temperatures has drastic consequences for the natural world. Thermal adaptation involves finding a match between temperatures permitting growth and the expected temperature distribution of the environment. However, if and how this match is achieved, and how tightly linked species change together, is poorly understood. Paramecium bursaria is a ciliate that has a tight physiological interaction with endosymbiotic green algae (zoochlorellae). We subjected a wild population of P. bursaria to a cold and warm climate (20 and 32℃) for ∼300 generations. We then measured the thermal performance curve (TPC) for intrinsic rate of growth (rmax) for these evolved lines across temperatures. We also evaluated number and size of the zoochlorellae populations within paramecia cells. TPCs for warm‐adapted populations were shallower and broader than TPCs of cold‐adapted populations, indicating that the warm populations adapted by moving along a thermal generalist/specialist trade off rather than right‐shifting the TPC. Zoochlorellae populations within cold‐adapted paramecia had fewer and larger zoochlorellae than hot‐adapted paramecia, indicating phenotypic shifts in the endosymbiont accompany thermal adaptation in the host. Our results provide new and novel insight into how species involved in complex interactions will be affected by continuing increasing global temperatures.

Rapid evolution of unimodal but not of linear thermal performance curves inDaphnia magna

“…The here observed evolution of a weaker plastic response may be beneficial as it resulted in a higher performance at temperatures beyond T opt compared with the control Daphnia, even though R max did not evolve. A similar adaptive decrease in thermal plasticity beyond T opt has also been found for population growth and photosynthesis in warm-adapted populations of two diatoms, a ciliate and a coral species [29,31,61]. While a set of life-history variables (survival, intrinsic population growth rate and fecundity at first clutch) had a unimodal TPC, other life-history variables (development rate, somatic growth rate and body mass) and the variables related to energy gain (ingestion rate) and costs (metabolic rate) had a linear TPC over the temperature range measured (note that all variables were log-scaled).…”

“…The here observed evolution of a weaker plastic response may be beneficial as it resulted in a higher performance at temperatures beyond T opt compared with the control Daphnia , even though R max did not evolve. A similar adaptive decrease in thermal plasticity beyond T opt has also been found for population growth and photosynthesis in warm-adapted populations of two diatoms, a ciliate and a coral species [29,31,61].…”

“…the ‘generalist–specialist hypothesis’) [29–31]. While this trade-off has been more typically associated with adapting to variable thermal regimes rather than a constant elevated mean, it has been found under the latter thermal conditions [29,31]. Therefore, the absence of the generalist–specialist trade-off in our study did not necessarily result from the similar amplitudes of temperature fluctuations that both selection groups experienced in the outdoor mesocosms.…”

“…Despite the evolution of the curvature of the TPCs, we could not detect support for the assumed trade-off between a broader thermal tolerance range and a decrease in R max (i.e. the ‘generalist–specialist hypothesis’) [29–31]. While this trade-off has been more typically associated with adapting to variable thermal regimes rather than a constant elevated mean, it has been found under the latter thermal conditions [29,31].…”

“…Here, we extend these studies by explicitly considering the evolution of TPCs, thereby reconstructing the evolution of thermal plasticity of multiple traits across a range of temperatures. We expected that, for log-scaled traits with a unimodal TPC, the heat-selected D. magna would have evolved (i) a higher T opt [18,21,22], which could be accompanied by a higher R max [17,29], or (ii) a larger thermal tolerance range (less curvature), which could be accompanied by a lower R max [29][30][31]. For log-scaled traits with a linear TPC, we expected the change of the activation energies (i.e.…”