Fig 1.
Schematic figure showing how 2-D and 3-D nutritional geometry approaches can identify strategies for prioritizing competing nutritional needs.
(A) In 2-D choice experiments, animals reveal their intake target as the nutritional blend that maximizes performance (blue dot). Here, an animal has consumed four successive meals, switching between carbohydrate-biased and protein-biased diets to reach a slightly carbohydrate-biased intake target (adapted from [19]). (B) In 2-D no-choice experiments, animals reveal a rule of compromise intake array when confined to single nutritionally imbalanced diets that constrain their intake P:C ratio to ‘nutritional rails’ extending from the origin. Here, colonies were confined to 5 P:C diets and cumulative intake values were measured for each diet treatment (orange dots). Dashed lines connect these intake values and reflect decisions about over-harvesting one macronutrient to avoid under-harvesting a more crucial limiting macronutrient relative to the intake target (blue dot). Generalist consumers tend to obey the ‘equal distance rule’, with a straight-line array, as they expect to redress a temporary imbalance by switching to another complementary food later. Specialist consumers tend to obey the ‘closest distance rule’, with a convex array reflecting efforts to stay as close to the intake target as possible (adapted from [7]). (C) Using 3-D nutritional landscapes, a fundamental macronutrient niche can be visualized with three nutrient mixtures shown in bivariate plots [22]. Here, each black dot is a specific mixture of protein, carbohydrates, and lipids (P:C:L). Protein and carbohydrate values are plotted on X and Y axes respectively, and lipids are plotted as diagonal lines with negative slopes that intersect P:C:L points with the sum of the 3 nutrients adding to 100%. In this example, organisms are provided seven P:C:L diets (e.g. 10:10:80 has 10% P + 10% C + 80% L) in a no-choice experiment. Diet harvest and performance are then mapped by interpolating between values measured at each diet (adapted from [23]). (D) We can define the fundamental macronutrient niche (FMN) by mapping regions of maximal diet harvest and/or consumption across the P:C:L landscape (adapted from [12]).
Fig 2.
Nutritional geometry in two dimensions.
In a choice experiment lasting 12 days, colonies reliably harvested and consumed a slightly carbohydrate-biased 1:1.5 P:C intake target in both choice diet combinations. In a no-choice experiment, colonies exhibited a generalist equal distance intake array (dashed lines) by harvesting similar excesses of protein and carbohydrates relative to their intake target. Diet consumption (subtracting scattered and hoarded amounts from harvested diet) tightly matched diet harvest. Solid black lines show nutrient rails for each no-choice P:C diet treatment. Choice experiment harvest values are provided with bi-directional error bars, and no-choice harvest values provided with pythagorean standard error bars aligned with the intake rail (as per [62]).
Fig 3.
Post-harvest diet processing (± SE) after 12 days in the 2-D choice experiment, based on analyzing uneaten colored diet.
(A) Colonies tended to preferentially hoard (retained inside the nest) carbohydrate-biased diets in both the 1:3 vs. 6:1 P:C pairing and the 1:6 vs. 3:1 P:C pairing treatments. (B) Colonies scattered diets evenly in the 1:3 vs. 6:1 P:C treatment, but scattered significantly more of the carbohydrate-biased diet in the 1:6 vs. 3:1 P:C treatment. Asterisks indicate the results of paired t-tests, testing difference within pairings between either hoarded or scattered diet (dry mass, mg), with n.s. = not significant, * p < 0.05, ** p < 0.01, *** p < 0.001).
Fig 4.
Nutritional landscapes visualize the fundamental macronutrient niche (FMN) of M. pharaonis in three dimensions (proteins, carbohydrates, and lipids; P:C:L).
(A) Colonies harvested maximal amounts across a broad range of protein-biased and carbohydrate-biased diets, while avoiding lipid-biased diets. (B) Colonies consumed most of the diet they harvested regardless of the P:C:L content. (C) The survival percentage of adult workers was highest on diets with moderate to high carbohydrates, low protein and low to moderate lipids. In contrast, moderate to high amounts of protein yielded the lowest worker survival, even as carbohydrates and lipids increased. (D) Colonies produced the most eggs when colonies were confined to diets with similar levels of protein and carbohydrates and low lipid content, and the fewest eggs on all diets where protein availability was low, regardless of carbohydrate and lipid amounts. Isoclines (red areas are highest values and blue areas are lowest) indicate dry mass of diet (mg), percent worker survival, or egg numbers, with scale bars adjusted relative to the range of observed values. Landscapes comprise 7 P:C:L ratios (see Fig 1C) and diet percentages (axis labels) indicate energy content provided by each macronutrient, with total energy content available standardized across diets (see Methods). Response surface regressions underlying colored heat maps were significant in each panel (Table 1).
Table 1.
Statistical output from Response Surface Models (RSM) analyses of 3-D feeding experiment.