The temperature-size rule (TSR) describes the inverse relationship between organism size and environmental temperature in uni- and multicellular species. Despite the TSR being widespread, the mechanisms for shrinking body size with warming remain elusive. Here, we experimentally test three hypotheses (differential development and growth [DDG], maintain aerobic scope and regulate oxygen supply [MASROS] and the supply-demand hypothesis [SD]) potentially  explaining the TSR using the aquatic protist Colpidium striatum in three gradually changing and one constant temperature environment crossed with three different nutrient levels.

We find that the constant and slowly warming environments show similar responses in terms of population dynamics, whereas populations with linear and fast warming quickly decline and  show a stronger temperature-size response. Our analyses suggest that acclimation may have played a role in observing these differences among treatments. The SD hypothesis is most parsimonious with the data, however, neither the DDG nor the MASROS hypothesis can be firmly dismissed. We conclude that the TSR is driven by multiple ecological and acclimatory responses, hence multicausal.

Data were collected from an aquatic microcosm experiment. The experiment has a full-factorial design of four levels of temperature treatment (constant 15°C, three rising temperature between 15-30°C and three levels of nutrient concentration (low = 0.28, medium = 0.56, and high = 1.12 g Protist Pellet medium per liter). We chose constant 15°C as the reference temperature as this is representative for the temperature Colpidium striatum experiences in natural freshwater environments. All three rising temperature treatments start at 15°C and end at 30°C. The linear speed is +1°C per day, while the fast nonlinear rising temperature increases at the beginning of the experiment (growth phase) faster then it slows down and the slow nonlinear rising temperature treatment goes in reversed fashion.

During the experiment, 5 mL of medium (of a total of 100 mL medium) was taken every day and replaced with 5 mL of sterile medium. 1 mL of the sample was transferred to a counting chamber and filmed under a stereo microscope with a 25× magnification mounted digital CMOS camera (Hamamatsu Orca C11440; Hamamatsu Photonics, Japan) to take a 5s video (25 frames per second). The following properties were measured with the R package BEMOVI: individual morphology (cell size and shape), and population size (individuals per mL). Furthermore, samples were processed to measure bacterial abundance using flow cytometry with an Accuri C6 flow cytometer and dissolved oxygen (DO) concentration with oxygen sensors (from PreSens - Precision Sensing GmbH); for details refer to Pennekamp et al 2015.

Missing data is coded as NA.