Using nutritional geometry to define the fundamental macronutrient niche of the widespread invasive ant Monomorium pharaonis

The emerging field of nutritional geometry (NG) provides powerful new approaches to test whether and how organisms prioritize specific nutritional blends when consuming chemically complex foods. NG approaches can thus help move beyond food-level estimates of diet breadth to predict invasive success, for instance by revealing narrow nutritional niches if broad diets are actually composed of nutritionally similar foods. We used two NG paradigms to provide different, but complementary insights into nutrient regulation strategies and test a hypothesis of extreme nutritional generalism in colony propagules of the globally distributed invasive ant Monomorium pharaonis. First, in two dimensions (protein:carbohydrates; P:C), M. pharaonis colonies consistently defended a slightly carbohydrate-biased intake target, while using a generalist equal-distance strategy of collectively overharvesting both protein and carbohydrates to reach this target when confined to imbalanced P:C diets. Second, a recently developed right-angled mixture triangle method enabled us to define the fundamental niche breadth in three dimensions (protein:carbohydrates:lipid, P:C:L). We found that colonies navigated the P:C:L landscape, in part, to mediate a tradeoff between worker survival (maximized on high-carbohydrate diets) and brood production (maximized on high-protein diets). Colonies further appeared unable to avoid this tradeoff by consuming extra lipids when the other nutrients were limiting. Colonies also did not rely on nutrient regulation inside their nests, as they did not hoard or scatter fractions of harvested diets to adjust the nutritional blends they consumed. These complementary NG approaches highlight that even the most successful invasive species with broad fundamental macronutrient niches must navigate complex multidimensional nutritional landscapes to acquire limiting macronutrients and overcome developmental constraints as small propagules.

Colonies relocated to defined nesting areas within hours of placement in arenas, which consisted of a black cardboard rectangle (3.5 x 2.5 cm) elevated slightly by a folded corner in the 2-D experiment, or a 4 cm 2 square with a small entrance, and a transparent red cellophane square covered by glass in the 3-D experiment. Lids containing pre-weighed experimental foods were placed in the foraging area near the nest. In the 2-D experiment, colonies were provided a fresh block (ca. 1 cm 3 ) of demineralized water mixed with agar (16g/L) each day. Agar-water cubes (16 g/L) were used since we expected workers to scatter bits of cotton from traditional cottonplugged water tubes in foraging areas, that may have hampered recovery of scattered diet (see below). However, we switched back to water tubes for the 3-D experiment when it became clear that scattered diet could be easily distinguished from small discarded cotton bits. To provide water, colonies were provided a fresh block (ca. 1 cm 3 ) of demineralized water mixed with agar (16g/L) each day (2-D experiment), or water tubes (3-D experiment).

Diet preparation
Diets were prepared using modified versions of a published protein:carbohydrate (P:C) diet [3] and a protein:carbohydrate:lipid (P:C:L) diet [6]. The P:C diet contained protein from whey powder (Myopure), whole egg powder (Great American Spice Company), and calcium caseinate (Arla), carbohydrates from sucrose (Sigma Aldrich), and micronutrients from Vanderzant vitamin mixture (Sigma Aldrich). Sucrose was used as the carbohydrate source, and dried egg white powder, whey protein and calcium caseinate were used as the protein source in approximately 1:1:1 ratio. The P:C:L diet used the same ingredients, but replaced whole-egg powder with egg-white only powder (Myoprotein), and provided lipids with a 4:1:1:1:1 ratio of lard:fish-oil:sunflower-oil:rapeseed-oil:peanut-oil. This mixture was chosen based on a series of pilot experiments showing that workers recruited most to lard, followed by similar recruitment levels to the 4 selected oils relative to 3 other less preferred oils (safflower oil, olive oil, coconut oil). Lard was melted, mixed with the other oils, and then combined with 2 ml of chloroform.
This mixture was combined with the dry ingredients and the chloroform was allowed to evaporate under a fume hood at room temperature for 96 hr (as per [6]). Diet recipes are provided in Table S1 and Table S2.
To prepare diets, agar was gently heated while stirring in water (16 g/L) until boiling and was then cooled slightly before being mixed in a blender with pre-weighed macronutrient ingredients. Mixed diets were then poured into petri dishes, sealed with parafilm, and stored at 4° C until provided to ants. For all experiments, we measured diet harvest by placing pre-weighed (initial wet mass) diet cubes (ca. 1 cm 3 ) on small dishes inside colony foraging areas, and then collecting them after 24 hours. These diets were then oven-dried at 60°C for 24 hours and weighed to the nearest 1 µg (final dry mass) on an AG285 Mettler Toledo microbalance. Each day, we also recorded wet and dry mass of 4 control cubes of each P:C diet, which enabled us to calculate dry:wet conversion factors used to estimate initial dry mass of each experimental diet cube [7]. We summed these diet values to calculate cumulative harvest for the experiment.
We also added a few drops of food coloring (Dr. Oetker TM ) to each diet just prior to blending all ingredients, which enabled us to separate it from debris when collecting hoarded (piled within the defined nest area) and scattered (discarded in the foraging area) diet at the end of the experiment. In the 2-D choice experiment, we further colored the two diets differently so that we could distinguish them to reconstruct macronutrient compositions in subsequent analyses. While diet color did not impact colony foraging behavior in pilot experiments, we nonetheless varied the color-nutrient combinations (i.e. the 1:6 diet was blue and 3:1 diet was red in one pairing and reversed in another pairing). Hoarded and scattered diets were oven dried and then weighed to the nearest 1 µg. Consumed diet was harvested diet minus the summed mass of hoarded and scattered diet.

2-D nutritional geometry experiment
Each colony was established with 200 workers and a scoop (0.5 x 0.5 x 0.15 cm) of brood, so that each colony had brood on day 1 of the experiment, and similar amounts of older brood which are known to be important for nutrient processing in M. pharaonis colonies [8,9].
Dead workers were collected from each colony during the four day acclimation period and were replaced on day 1 of the experiment to standardize initial colony size. Queens were not included in this experiment because we were interested in colony foraging decisions rather than colony growth performance. While M. pharaonis workers, lacking an egg laying queen, may have attempted to convert eggs or first instar larvae present in the initial scoop of brood into sexuals (Pontieri, pers. comm.), which may have altered their foraging behavior, we did not detect any sexual larvae or pupae in colonies during the 12-day experiment. We assigned 24 colonies to the choice experiment (n = 12 colonies per choice pairing treatment, but removed one colony from each choice treatment due to missing intake data on one day), leaving 11 colonies per choice pairing treatment. We also assigned 40 colonies to the no-choice experiment (n = 8 colonies per diet treatment). Over 12 days in both choice and no-choice experiments, we replaced old diet with fresh diet daily. We also counted and collected dead workers every fourth day during the experiment and collected the remaining living workers on day 12.

3-D nutritional geometry experiment
The P:C:L diet treatments did not significantly affect the number of larvae (explained 2% of variation in this response variable), although they explained a significant amount (23%) of variation in egg number (Table 1). This difference was likely due to larva number representing brood present during the acclimation phase (when all colonies received the same standard diet), but egg number representing queen health and activity during the colonies' exposure to P:C:L diet treatments. Thus, larva number was not considered further in the study. Additionally, P:C:L landscapes were not produced for scattered or hoarded diet results, since the overall RSM models were also not significant for these variables (Table 1).