Workers readily picked up the offered leaf discs and transported them back into the nest. When all fungus chambers were well nourished, there was no difference in the distribution pattern over the day and all chambers were equally supplied (Fig. 2a, for the sake of uniformity, data are presented as box plots, as data of the ‘nourished’ experiment met normality and equal variance, while the pooled data of the ‘undernourished’ experiment did not; ‘nourished’ experiment: 2-way repeated-measures ANOVA, factor time of day P = 0.275, factor chamber P = 0.376).
An even resource distribution was no longer the case when one of the fungus gardens experienced undernourishment. While time of day had also no effect on the distribution pattern as in the previous experiment, chambers were supplied differently (Fig. 2a, 2-way repeated-measures ANOVA, factor time of day P = 0.61, factor chamber P < 0.001). The undernourished chamber 2 received almost twice the number of discs compared to chamber 1 (medianchamber 1 = 30, medianchamber 2 = 61), while chamber 3 (medianchamber 3 = 19) received much less (Fig. 2a, pooled data, time of day 10:00, 12:00, and 14:00 h; Kruskal–Wallis test P < 0.001; Tukey post hoc test, all chambers pairwise comparisons, P < 0.05). It appears likely that the reduced delivery to chamber 3 resulted from the limited number of discs offered in each assay (100 discs), rather than from a workers’ response aimed at avoiding such chamber, since a higher delivery to one chamber would logically leave less discs to be delivered to another chamber.
Supply of chamber 2 did not lessen 1 day later, and a pattern similar to day 1 could be observed (Fig. 2a, 2-way repeated-measures ANOVA, factor time of day P = 0.62, factor chamber P < 0.01; pooled data, time of day 10:00, 12:00, and 14:00 h; Kruskal–Wallis test P < 0.001; Tukey post hoc test, all chambers pairwise comparisons, P < 0.05). The similar distribution pattern of day 1 and day 2 after leaf deprivation could either result from an enduring demand of the fungus, as leaf fragments collected the first day and overnight had not mitigated undernourishment, or because workers continued to respond to orientation cues although the fungus may have recovered. Overall, our results indicate that LCA can perceive the undernourished condition of their fungus and specifically react with a higher supply. In the ant Temnothorax albipennis, Solenopsis invicta, and Lasius niger, workers can also perceive starvation in nestmates and brood, and react with increased liquid uptake (Howard and Tschinkel 1980) or by supplying the same number of workers, or even more workers at a faster speed (Sendova-Franks et al. 2010; Mailleux et al. 2011).
In the ‘nourished’ experiment, most workers delivered the very first of the offered leaf discs to chamber 1, the closest to the nest entrance (Fig. 2b). Chambers 2 and 3 rarely received the first delivery (statistics: Table 1a). In the ‘undernourished’ experiment, however, while chamber 1 still received most of the first delivered discs on day 1 as well as on day 2, clearly more workers chose the undernourished chamber 2 for their first delivery. Statistically significant differences resulted from a change of highest first delivery from chamber 1 (nourished experiment) to chamber 2 (undernourished experiment), while chamber 3 only rarely received the first delivery of the offered leaf discs in either experiment (Table 1b).
Table 1 Statistics for the first delivery of offered leaf discs, after Fisher’s exact test: a Pairwise comparisons of the overall distribution pattern between different experimental conditions. For each chamber, counts of delivered discs were pooled for the three time points, as there were no statistical differences in the delivery among them; b Pairwise comparisons between different experimental conditions, for each chamber Do ants recognize the nutritional state of a fungus garden without entering the chamber? If so, ants should perceive and respond to putative volatiles emanating from the fungus garden via the side tunnel. An undernourished fungus could emit a volatile blend that differs from the emission of a symbiont in a good nutritional condition. Alternatively, leaf supply to chambers could be guided by pheromone markings. Laden LCA mark their foraging trails on their way back to the nest (Robinson et al. 1974). They could continue these markings beyond the nest entrance until they reach a chamber. Other laden workers could follow that trail into the chamber while also laying a pheromone trail, reinforcing the supply by positive feedback until demand in this chamber is satisfied. As the undernourished chamber had a higher demand, trail laying could have lasted longer, attracting more foragers and thus increasing the traffic flow towards that chamber. Future analysis of ant traffic flows under different nourishment levels could help to uncover the involvement of chemical trails in the organization of intra-nest resource distribution.
Interestingly, more foragers bypassed chamber 1 and delivered the first of the offered leaf discs to chamber 2 when that fungus garden was undernourished. This hints at the existence of informed foragers stemming from chamber 2 that supplied leaf discs directly to the fungus garden they originated from. In addition, the increased overall delivery to the undernourished garden could also be caused by the involvement of more workers originating from that chamber, likely with lower foraging thresholds, going out and returning to it loaded with a disc. Importantly, these foragers were not undernourished themselves, as they received diluted honey ad libitum during the undernourishment of their garden, so that they appear to respond directly to the state of the fungus. In fact, Acromyrmex lundii LCA are known to respond to so far unknown cues from the fungus garden, and to discontinue foraging of leaf materials previously experienced as harmful to the fungus but not to themselves (Herz et al. 2008).
LCA nests may house up to eight thousand fungus chambers (Moreira et al. 2004) that need to be supplied with leaf fragments. In large nests, there can be quite some distance between chambers situated at different locations in the network. Despite this, LCA are able to evenly distribute resources within the nest (Pretto and Forti 2000), so that any distance effect on resource distribution, as observed in the first delivery of offered fragments in our experiment, was transitory and no longer noticeable over time. We propose that overprovisioning of the chambers closest to the nest entrance is avoided via the following mechanisms, which ultimately may result in an even resource distribution: (i) a high delivery to a first chamber may generate congestion of loaded workers at the chamber entrance, and prompt some of them to continue walking downstream to the next chamber, and so a; (ii) loaded foragers may, after having entered a well-provisioned chamber, leave it without unloading and walk downstream; (iii) inside-nest workers may remove and relocate some fragments to the next downstream chambers. In Atta colombica LCA, it was observed that fungus gardens located farther from the nest entrance along a series of three chambers were supplied with large delays (Burd and Howard 2005), as expected because of the distance effect. However, the three nest chambers were serially yet directly connected one after another, so that leaf fragments transported downstream had to be carried through the previous chambers, a situation that is not observed in natural nests. Consequently, not only the distance but also the size of the carried fragments was reported to affect the speed of their transport, which is likely an experimental artifact of the unnatural nest arrangement used. Therefore, the hypothesis that the constraints imposed by the underground transport and processing of leaf fragments may have shaped the evolution of fragment size determination in LCA, as advanced by the authors, remains elusive.
As avenues for future research, we hypothesize that LCA achieve a suitable leaf supply to their fungus chambers via the following behavioral mechanisms: (1) informed workers guide supply to chambers with higher resource demand by laying pheromone trails; (2) naïve workers orient towards volatiles emanating from fungus chambers and/or follow pheromone trails laid within the nest; (3) overprovisioning is prevented by relocation of excessive leaf fragments to nest chambers with higher demand, or by temporary storage in leaf caches, from where they are delivered to neighboring fungus chambers.