Experimental setup. (A) No-choice experiment, diet 1:2, with a total concentration of 40 g·L⁻¹. (B) Choice experiment, diet 6:1 vs. diet 1:2. (C) Multiple-choice experiment. The slime mold was initially placed at the center of the petri dish.

“Intake” regulation when faced with two foods varying in protein (P) and carbohydrate (C) content. Mean growth rate [initial mass (mg)/final mass (mg)] ± SD as a function of the mean proportion of carbohydrates ingested ± SD [C intake/(P intake + C intake)], measured for the binary choice experiment (SI Text describes method used to estimate intake) (n = 10 slime mold fragments per choice treatment, total dietary concentration of P and C in each food was 80 g·L⁻¹). The black curve comes from the no-choice experiments (Fig. 2C).

Performance responses. Data were recorded for individual slime molds confined for 60 h to 1 of 35 diets varying in both the ratio and total amount of protein and carbohydrate (40 g·L⁻¹, 9 ratios; 80 g·L⁻¹, 17 ratios; and 160 g·L⁻¹, 9 ratios). Response surfaces were visualized using nonparametric thin-plate splines, which were fitted using the fields package (National Center for Atmospheric Research, Boulder, CO) in the statistical software R (36). Red indicates the highest values for the experimental variable on a given response surface, with values descending to lowest values in dark blue regions. (A) Effect of carbohydrate (C) concentration on proportion of slime that survived. The logistic regression analysis yielded a significant relationship between survival and carbohydrate concentration (χ² = 299.81, P < 0.001; z = 18.91, P < 0.001). (B) Effects of diet composition on slime mold density [final mass (mg)/final area (cm²)]. The slime molds that did not survive were not included. (C) Effects of diet composition on growth rate [initial mass (mg)/final mass (mg)]. (D) Mean growth rate on a mass basis ± SD as a function of the proportion of carbohydrate in the diet: C/(P + C). The polynomial regression analysis yielded a significant relationship between growth rate and proportion of carbohydrate in the diet (R² = 0.90, F_2,14 = 63.46, P < 0.001). (E) Effects of diet composition on expansion rate [(initial area (cm²)/final area (cm²)]. The response surface regression analyses yielded significant relationships as follows: R² = 0.65, F_5,344 = 125.54, P < 0.001 for survival (surface not plotted, Table S1); R² = 0.50, F_5,294 = 57.74, P < 0.001 for density (Fig. 2B and Table S2); R² = 0.50, F_5,344 = 69.80, P < 0.001 for growth rate (Fig. 2C and Table S1), and R² = 0.54, F_5,344 = 80.44, P < 0.001 for expansion rate (Fig. 2E and Table S1).

“Intake” regulation when faced with multiple foods varying in protein and carbohydrate content. (A) Observed proportional coverage by slime molds of the different foods in the 11-food array compared with the expected proportion if slime molds had grown to cover foods indiscriminately (n = 30 slime molds). To test whether slime molds preferred particular nutrient ratios over others, we used a binomial test on the number of slime molds covering each diet. The null hypothesis was that slime molds could cover each food with equal probability (we computed the probability to cover a particular food knowing that one slime mold could cover between 1 and 3 foods at a time but not more). Slime molds significantly avoided extremely carbohydrate (C)-biased diets (binomial test: P < 0.05 for 1:4, 1:6, and 1:9 ratios) and preferred diets that contained equal proportions of protein (P) and carbohydrate or were slightly protein-biased (binomial test: P < 0.05 for 1:1 and 2:1 ratios). (B) Mean growth rate in term of mass ± SD as a function of the mean proportion of carbohydrates ingested ± SD [C intake/(P intake + C intake)], measured for the multiple-choice experiment (n = 30 slime molds, 11 foods all with a macronutrient concentration of 80 g·L⁻¹). The black curve comes from the no-choice experiments (Fig. 2C).