**Background:**
Animal contest theory predicts that the level of escalation in contests should increase with the value of the contested resource. While this has been empirically supported in dyadic contests (e.g., males competing for mates, females for oviposition sites), it has not been experimentally tested in group contests. Testing this in group-living animals is challenging because resource value can differ between contestants based on physiological state and prior experience, and individual differences may obscure group-level patterns. The Geometric Framework for nutrition offers a tool to manipulate resource value experimentally by altering dietary history, as individuals and colonies compensate their foraging to meet nutritional targets. The authors used the Australian meat ant Iridomyrmex purpureus, whose neighbouring colonies maintain display grounds where workers engage in typically non-injurious agonistic displays, though lethal fights can occur over food resources such as honeydew-secreting hemipterans.
**Methods:**
Field experiments were conducted at Serendip Sanctuary, Victoria, over two summers (2020–2021). First, compensatory foraging assays (November 2019) tested whether colonies adjust foraging effort based on prior diet. Fourteen colony pairs were pre-fed either carbohydrate (50 g of 25% honey solution) or protein (40 g of canned tuna in springwater) for five days. On day six, colonies within each group were randomly assigned to receive either honey or tuna, and the number of workers at feeders was counted at 30-minute intervals. Results showed that colonies pre-fed honey deployed significantly more workers to tuna than colonies pre-fed tuna (W = 49, p < 0.001), but pre-feeding did not affect deployment to honey (W = 12, p = 0.13). This established that tuna value could be manipulated: high value (pre-fed honey) vs. low value (pre-fed tuna).
For staged contests, 55 colony pairs with active display grounds were used. After the same pre-feeding protocol and a compensatory foraging test, 30 g of tuna was placed on the display ground. After 20 minutes, a 30×30 cm quadrat was placed above the food, and digital images were taken to count total competing workers. The number of workers engaged in physical 'grappling' (often fatal) was counted over one minute. Statistical analyses used Linear Models and binomial GLMs, with food value, compromised feeding (accounting for testing-phase diet differences), and season (early vs. late summer) as fixed effects.
**Key Results:**
Compensatory foraging was consistent across experiments: colonies pre-fed honey deployed significantly more workers to tuna than those pre-fed tuna (t = −7.93, p < 0.001; N = 70), while deployment to honey was unaffected by pre-feeding (t = −0.65, p = 0.52; N = 70). In staged contests, colonies deployed significantly more workers to contested tuna when it was high-value (pre-fed honey) vs. low-value (pre-fed tuna) (t = −4.82, p < 0.001; N = 55), a pattern consistent across seasons but more pronounced in early summer (effect of late season: t = −4.23, p < 0.001). Grappling occurred in 33% of contests (18 out of 55). The proportion of contests escalating to grapples was significantly higher for high-value food than low-value food (Z = −3.15, p = 0.002), and this did not differ by season (Z = 1.28, p = 0.20). The probability of escalation depended on worker number (Z = 2.65, p = 0.008). Among contests with grappling (N = 18), the proportion of grappling ants did not vary with worker number (F₁,₁₄ = 1.20, p = 0.29), season (F₁,₁₄ = 0.22, p = 0.65), or food value (F₁,₁₄ = 0.003, p = 0.95). The proportion of ants grappling was low (≤ 0.06).
**Clinical Implications:**
This study is not a clinical investigation and has no direct clinical implications. However, it provides a robust experimental framework for testing contest theory in group-living animals, demonstrating that colony-level nutritional demands drive escalation behaviour rather than individual worker physiology. The Geometric Framework approach avoids confounding factors such as individual motivation or metabolic state that arise from food deprivation protocols. This methodology could be applied to other field studies of both dyadic and group contests across taxa.